Вы находитесь на странице: 1из 197


Relativity for Poets

Benjamin Crowell

Fullerton, California

Copyright 2015
Benjamin Crowell

rev. December 5, 2015

Permission is granted to copy, distribute and/or modify this document under the terms of the Creative Commons Attribution
Share-Alike License, which can be found at creativecommons.org.
The license applies to the entire text of this book, plus all the
illustrations that are by Benjamin Crowell. All the illustrations
are by Benjamin Crowell except as noted in the photo credits
or in parentheses in the caption of the figure. This book can
be downloaded free of charge from www.lightandmatter.com in
a variety of formats, including editable formats.


Space and time

1 Galilean relativity
1.1 Galileo versus Aristotle . . . . . . . .
1.2 Frames of reference . . . . . . . . . . .
1.3 Inertial and noninertial frames . . . .
1.4 Circular motion . . . . . . . . . . . . .
1.5 Addition of velocities . . . . . . . . . .
1.6 The Galilean twin paradox . . . . . .
1.7 The Galilean transformation . . . . . .
1.8 Science and society: The Galileo affair
2 Einsteins relativity
2.1 Time is relative . . . . . . . .
2.2 The principle of greatest time
2.3 A universal speed limit . . . .
2.4 Einsteins train . . . . . . . .
2.5 Velocities dont simply add .
2.6 The light cone . . . . . . . . .
2.7 Science and society: Scientific
3 The

Lorentz transformation
Surveying space and time .
The Lorentz transformation
Correspondence principle .
Scale, squish, stretch . . . .




. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .



























No acceleration past c . . . . .
Time dilation . . . . . . . . . .
Length contraction . . . . . . .
Not what you see . . . . . . . .
Causality . . . . . . . . . . . .
Doppler shift . . . . . . . . . .
Optional: Proof that area stays

. .
. .
. .
. .
. .
. .

. . . .
. . . .
. . . .
. . . .
. . . .
. . . .







4 The

measure of all things

Operationalism . . . . . . . . .
Galilean scalars and vectors . .
Relativistic vectors and scalars
Measuring relativistic vectors .














Matter and E = mc2


5 Light and matter

5.1 How much does matter really matter? . . . . . . . 71
5.2 How hard, how loud? . . . . . . . . . . . . . . . . . 72
5.3 Light . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6 E = mc2
6.1 Energy shift of a beam of light . .
6.2 Einsteins argument . . . . . . . .
6.3 Correspondence principle . . . . .
6.4 Mass to energy and energy to mass
6.5 The energy-momentum vector . . .
6.6 Tachyons . . . . . . . . . . . . . .














7 Newtonian gravity
7.1 Galileo and free fall . . . . . . . . . . . . . . . . . . 93
7.2 Orbits . . . . . . . . . . . . . . . . . . . . . . . . . 96
7.3 Newtons laws . . . . . . . . . . . . . . . . . . . . . 98

8 The

Tidal forces . . . . . . . . . . . . . . . . . . . . . . 100

equivalence principle
The equivalence principle .
Chicken and egg . . . . . .
Apparent weightlessness . .
Locality . . . . . . . . . . .
Gravitational time dilation
Time dilation with signals .

9 Curvature of spacetime
9.1 What is straight? . . . .
9.2 The parable of the bugs
9.3 Gravity as curvature . .
9.4 Deflection of starlight .
9.5 No extra dimensions . .
9.6 Parallel transport . . . .































10 Matter
10.1 No nongravitating matter . . . . . . . . . . . . .
10.2 No repulsive gravity . . . . . . . . . . . . . . . .
10.3 Finite strength of materials . . . . . . . . . . . .

. 130
. 132
. 134

11 Black holes
11.1 Collapse to a singularity . .
11.2 The event horizon . . . . .
11.3 Detection . . . . . . . . . .
11.4 Light cones . . . . . . . . .
11.5 Penrose diagrams . . . . . .
11.6 Tests of general relativity in
11.7 Speed of light . . . . . . . .

. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
weaker fields
. . . . . . . .







12 Waves
12.1 No action at a distance
12.2 Fields . . . . . . . . . .
12.3 Electromagnetic waves .
12.4 Gravitational waves . .




















13 For dust thou art

13.1 The Einstein field equation . . . .
13.2 Conservation . . . . . . . . . . . .
13.3 Change or changelessness? . . . . .
13.4 Homogeneity and isotropy . . . . .
13.5 Hubbles law . . . . . . . . . . . .
13.6 The cosmic microwave background
13.7 A hot big bang . . . . . . . . . . .
13.8 A singularity at the big bang . . .
13.9 A cosmic calendar . . . . . . . . .











14 Dark matter and dark energy

14.1 A difficult census . . . . . . . . . . . . . . . . . . . 185
14.2 Dark matter . . . . . . . . . . . . . . . . . . . . . . 186
14.3 Dark energy . . . . . . . . . . . . . . . . . . . . . . 188

Part I

Space and time

Chapter 1

Galilean relativity

Galileo versus Aristotle

Once upon a time, there was an ornery man who liked to argue.
He was undiplomatic and had a talent for converting allies into
enemies. His name was Galileo, and it was his singular misfortune to be correct in most of his opinions. When he was arguing,
Galileo had a few annoying habits. One was to answer a perfectly
sound theoretical argument with a contradictory experiment or
observation. Another was that, since he knew he was right, he
freely made up descriptions of experiments that he hadnt actually done.
Galileos true opponent was a dead man, the ancient Greek
philosopher Aristotle. Aristotle, probably based on generalizations from everyday experience, had come up with some seemingly common-sense theories about motion.



The figure on the facing page shows an example of the kind

of observation that might lead you to the same conclusions as
Aristotle. It is a series of snapshots in the motion of a rolling
ball. Time moves forward as we go up the page. Because the
ball slows down and eventually stops, it traces out the shape of
a letter J. If you get rid of the artistic details, and connect the
drawings of the ball with a smooth curve, you get a graph, with
position and time on the horizontal and vertical axes. (If we had
wanted to, we could have interchanged the axes or reversed either
or both of them. Making the time axis vertical, and making the
top point toward the future, is a standard convention in relativity,
like the conventional orientation of the compass directions on a
Once the ball has stopped completely, the graph becomes a
vertical line. Time continues to flow, but the balls position is
no longer changing. According to Aristotle, this is the natural
behavior of any material object. The ball may move because
someone kicks it, but once the force stops, the motion goes away.
In other words, vertical lines on these graphs are special. They
represent a natural, universal state of rest.
Galileo adduced two main arguments against the Aristotelian
view. First, he said that there was a type of force, friction, which
was the reason that things slowed down. The grass makes a
frictional force on the ball. If we let the grass grow too high,
this frictional force gets bigger, and the ball decelerates more
quickly. If we mow the grass, the opposite happens. Galileo
did experiments in which he rolled a smooth brass ball on a
smooth ramp, in order to make the frictional force as small as
possible. He carried out careful quantitative observations of the
balls motion, with time measured using a primitive water clock.
By taking measurements under a variety of conditions, he showed
by extrapolation from his data that if the ramp was perfectly
level, and friction completely absent, the ball would roll forever.
Today we have easier ways to convince ourselves of the same
conclusion. For example, youve seen a puck glide frictionlessly
across an air hockey table without slowing down. Sometimes





Galileos analysis may be hard to accept. Running takes us a lot

of effort, and it seems as though we need to apply a continuous
force to keep going. The figure above shows whats really happening. At b, the runners heel strikes the ground, and friction
slows him down. A forward force in c serves only to recover from
the loss in speed.
Galileos second argument had to do with the fact that we
cant judge motion except by comparing with some reference
point, which we consider to be stationary. Aristotle used the
earth as a reference point. But in 1610, Galileo looked at the
planet Jupiter through a telescope that he had invented, and
saw something startling. The planet, which looks like a bright
and starlike point to the naked eye, appeared as a disk, and accompanying it were four points of light. Observing them from
one night to the next, as shown in the notebook page reproduced
on the facing page, he saw that they wove their way back and
forth, to the east and west of Jupiters disk, in the plane of the
solar system. He inferred that they were moons that were circling Jupiter, and that he was seeing the circles from the side, so
that they appeared flat. (The analogy with our diagram of the
rolling ball is nearly exact, except that Galileo worked his way
down the page from night to night, rather than going up.)
The behavior of these moons is nothing like what Aristotle would have predicted. The earth plays no special role here.
The moons circle Jupiter, not the earth. Galileos observations
demoted the earth from its Aristotelian place as the universal
reference point for the rest of the cosmos, making it into just one
of several planets in the solar system.






Frames of reference

If Aristotle had grown up on Jupiter, maybe he would have considered Jupiter to be the natural reference point that was obviously at rest. In fact there is no one reference point that is always
right. Psychologically, we have a strong tendency to think of our
immediate surroundings as being at rest. For example, when you
fly on an airplane, you tend to forget that the cabin is hurtling
through the sky at nearly the speed of sound. You adopt the
frame of reference of the cabin. If you were to describe the motion of the drink cart progressing down the aisle with your nuts
and soda, you might say, Five minutes ago, it was at row 23,
and now its at row 38. Youve implicitly laid out a number
line along the length of the plane, with a reference point near
the front hatch (where row zero would be). The measurements
like 23 and 38 are called coordinates, and in order to define them
you need to pick a frame of reference.
In the top part of the figure on the next page, we adopt the
cows frame of reference. The cow sees itself as being at rest,
while the car drives by on the road. The version at the bottom
shows the same situation in the frame of the driver, who considers
herself to be at rest while the scenery rolls by.
Aristotle believed that there was one special, or preferred
frame of reference, which was the one attached to the earth.
Relativity says there is no such preferred frame.


Inertial and noninertial frames

Thats not to say that all frames of reference are equally good.
Its a little like the line from Orwells satirical novel Animal Farm:
All animals are equal, but some animals are more equal than
others. For an example of the distinction between good and
bad frames of reference, consider a low-budget movie in which
the director wants to depict an earthquake. The easy way to do
it is to shake the camera around but we can tell its fake.





Similarly, when a plane hits bad turbulence, its possible for

passengers to go popping out of their seats and hit their heads
on the ceiling. Thats why the crew tells you to buckle your
seatbelts. Normally we assume that when an object is at rest, it
will stay at rest unless a force acts on it. Furthermore, Galileo
realized that an object in motion will tend to stay in motion. This
tendency to keep on doing the same thing is called inertia. One of
our cues that the bouncing cabin is not a valid frame of reference
is that if we use the cabin as a reference point, objects do not
appear to behave inertially. The person who goes flying up and
hits the ceiling was not acted on by any force, and yet he changed
from being at rest to being in motion. We would interpret this
as meaning that the planes motion was not steady. The person
didnt really pop up and hit the ceiling. What really happened,
as suggested in the figure, was that the ceiling swerved downward
and hit the person on the head. The cabin is a noninertial frame
of reference. Moving frames are all right, but noninertial ones


Circular motion

In the era of Galileo and Newton, there was quite a bit of confusion about whether inertial motion was a broad enough notion
to include both motion in a straight line and motion in a circle.
Circular motion is not inertial. The figure is an overhead view
of a person whirling a rock on the end of a string. When the
string breaks, the rock flies off straight. It doesnt keep going
in a circle. The inward force from the string is necessary to the
circular motion, and when it disappears, we get straight-line motion. Similarly, when we spin a coin on a tabletop, the atoms



in the coin are being held together by attractive electrical forces

from the neighboring atoms.
Linear, inertial motion is never detectable without reference
to something external. Not so for circular motion. For example,
if the room in which youre reading this book started to rotate,
you would be able to tell because, for example, youd feel dizzy.
This would be your inner ear acting like a gyroscope. Passenger
jets use a type of optical gyroscope in order to sense and maintain
their orientation and stay on course. For these reasons, a rotating
frame of reference is not considered an inertial frame.


Addition of velocities

All inertial frames of reference are equally valid. However, observers in different frames of reference can give different answers
to questions and get different, correct answers from measurements. For example, suppose that Sparkly Elf Lady is riding her
unicorn at 30 kilometers per hour, and she fires an arrow from
her bow in the forwar direction. To her, the bow fires the arrow
at its normal speed of 20 km/hr. But to an observer watching
her go by the arrow is going 50 km/hr. To convert velocities
between different frames of reference, we add and subtract.



As a more realistic example, with real-world consequences,

Air France flight 447 disappeared without warning on June 1,
2009, over the Atlantic Ocean. All 232 people aboard were killed.
Investigators believe the disaster was triggered because the pilots
lost the ability to accurately determine their speed relative to the
air. This is done using sensors called Pitot tubes, mounted outside the plane on the wing. Automated radio signals showed that
these sensors gave conflicting readings before the crash, possibly
because they iced up. For fuel efficiency, modern passenger jets
fly at a very high altitude, but in the thin air they can only fly
within a very narrow range of speeds. If the speed is too low, the
plane stalls, and if its too high, it breaks up. If the pilots cant
tell what their airspeed is, they cant keep it in the safe range.

Many peoples reaction to this story is to wonder why planes

dont just use GPS to measure their speed. The answer has to
do with addition of velocities. Passenger jets do have GPS these
days, but GPS tells you your speed relative to the ground, not
relative to the air. We have

(velocity of the plane relative to the ground)

=(velocity of the plane relative to the air)
+(velocity of the air relative to the ground).

Given two of these numbers, you could find the third, but GPS
only tells you one of these numbers, and the one it tells you is
not the one that the pilots on Air France 447 needed to know.




The Galilean twin paradox

Alice and Betty are identical teenage twins. Betty goes on a

space voyage to get a fish taco with cilantro and grated carrots
from a taco shack that has the best food in the known universe.
Alice stays home. The diagram shows the story using our usual
graphical conventions, with time running vertically and space

Motion is relative, so it seems that it should be equally valid

to consider Betty and the spaceship as having been at rest the
whole time, while Alice and the planet earth traveled away from
the spaceship along line segment 3 and then returned via 4. But
this is not consistent with the experimental results, which show
that Betty undergoes a violent deceleration and reacceleration at
her turnaround point, while Alice and the other inhabitants of
the earth feel no such effect.
The paradox is resolved because Galilean relativity doesnt
say that all motion is relative. Only inertial motion is relative.
When motion is noninertial, we can get detectable physical effects. The earths motion is inertial, and the spaceships isnt.




The Galilean transformation

In section 1.2 on p. 16, we visualized the frames of reference of the

cow and the car by making two different diagrams of the motion,
one with the cow at rest and one with the car at rest. Rather
than making two separate graphs, we can be more economical
by drawing a single diagram, but superimposing two different
graph-paper grids, as shown above.
The white grid is what the cow would use. The car travels
one meter forward for every second of time.
The driver of the car would prefer the black grid. According
to this grid, the cow starts at t = 0 with a position of 3 meters,
and by t = 3 seconds, its gone backward to a position of 0 meters.
What were doing here is switching back and forth between
the coordinates of two different frames of reference. Coordinates
are like names that we attach to events. When we switch coordinate systems, reality doesnt change, only the names do. Its just
like translating words to a different language. This method of
translating between two frames of reference is called the Galilean




The Galileo affair

Galileo had a famous showdown with the Church, which ought to

teach us something useful about the relationship between science
and religion and their roles in society. But having taught about
this topic for twenty years, I find that the more I learn about it,
the less sure I am of what lessons to draw.
First some background. Galileos lifetime (1564-1642) coincided exactly with the Counter-Reformation, and during his mature years Europe was consumed by a period of brutal warfare
much like what we see today in places like Iraq, with religious
and state powers cynically using one another to further their own
agendas. In some areas, 75% of the population was killed by the
war and its side-effects, including disease and famine. In parallel
with the hard work of physical slaughter, there was an intellectual
battle going on. New religious orders such as the Theatines were
founded, with the mission of defending Italy from the Protestant heresy. In 1559 the Church published an Index of Forbidden
Books (abolished in 1966), and in 1588 the Roman Inquisition
was established. During the height of the conflict Galileo lived in
relative safety near Florence, with a Medici prince as his patron.
The security of his position was also seemingly buttressed because his scientific work was approved of by both the Jesuits and
his friend Cardinal Barberini, who in 1623 became Pope Urban
But Galileo was a flamboyant personality and a best-selling
author, and he ended up getting in trouble by very publicly advocating the Copernican system of the universe, in which the
planets, including the earth, circled the sun. After some initial
ignorant fulmination against Galileo from the pulpit, the first real
shot across the bow came in the form of an intellectually sophisticated 1616 letter to Galileo from the Theatine priest Ingoli, who



had participated in the Accademia dei Lincei, of which Galileo

was a leading member. Apologists for the Church love Ingolis
letter, because it zeroes in on two real scientific weaknesses in
Galileos position. This is contrary to the usual picture of the
Church persecuting poor Galileo purely because they thought a
moving earth contradicted passages from the Bible, e.g., Tremble before him, all the earth. The world is established and can
never be moved, (1 Chronicles 16:30). Although Ingoli does include such theological points, they are downplayed. Ingoli, basing his main arguments on work by the astronomer Tycho Brahe,
points out two purely scientific problems with Copernicanism:
1. Based on the best available data, the scale of a Copernican
universe would have had to be such that the stars were
implausibly large (about the size of our entire solar system).
2. If the earth were spinning on its own axis, then there would
be detectable effects on the motion of projectiles, but no
such effects are observed.
The first argument turns out to have been based on mistaken
data, but we can recognize the second as an argument about
relativity. Lets analyze this with the benefit of modern hindsight,
since neither Galileo nor Ingoli had a clear enough concept of
inertial motion to be able to attack it definitively.
Ones initial reaction might be that motion is relative, so if we
see the sun appear to spin around the earth, isnt it just a matter of opinion whether the earth is spinning or the sun revolving
around it? From this point of view, arguing about an earthcentered cosmos versus a sun-centered one is like arguing about
whether its really the cow or the car on p. 17 that moves. This
was in fact essentially the position taken by the higher Church
officials later on: they told Galileo that he could discuss Copernicanism as a mathematical hypothesis, but not as a matter of
physical fact.
But this argument is just plain wrong, since inertial motion
is not just motion at constant speed, its motion that is also



in a straight line. Therefore motion in a circle really is more

than a matter of opinion, and in fact we can detect the effects of
the earths rotation on projectiles. For example, when powerful
naval guns fire a shell to the north, in the northern hemisphere,
the shell can fly as far as 30 km, and its motion carries it to
a higher latitude, at which the earths rotation has a smaller
velocity. (As an extreme case, Santas candy-cane pole doesnt
go anywhere as the earth rotates, it just spins in place.) The
projectile retains its original, higher eastward velocity, so to an
observer on the earths surface, it appears to swerve to the east.
This is known as the Coriolis effect, and is also the explanation for
the rotation of cyclones. So Galileo was right about the earths
spin, but for the wrong reasons. At the time, the state of physics
simply wasnt advanced enough to accurately calculate the effect
Ingoli described. Such a calculation would have shown that the
effect was too small to have been observable with contemporary
The aged Galileo persisted stubbornly in advocating Copernicanism, and in 1633 he was sentenced to house arrest for the
rest of his life. Tradition holds that after the sentence was pronounced, he muttered, And yet, it does move.
So what do we learn from all this? A fairly common interpretation is that the whole conflict was needless. There doesnt have
to be any conflict between science and religion. We simply need
to keep science, religion, and the state within their separate and
proper spheres of authority. Although Ive given this interpretation to my students in the past, I cant help feeling now that I
was being a Pollyanna. As I write this, the press is reporting on
the killing of 12 journalists and 4 Jews by Muslim fundamentalists in Paris, as well as the massacre of 2,000 civilians in northern
Nigeria by Boko Haram, an Islamic fundamentalist group whose
name means literally books are a sin. Maybe a more realistic
moral to draw from the Galileo story is that religion is an inherently dangerous force in the world, one which tends to wreak
havoc in the secular world unless its backed into a corner with
a whip, like a snarling circus lion.



Chapter 2

Einsteins relativity

Time is relative

In chapter 1 I laid out a beautiful theory of motion, which, by

the way, is wrong. To see why its wrong, we have to think about
the assumptions hidden within it about space and time.
Suppose that I drive to Gettysburg, Pennsylvania, and stand
in front of the brass plaque that marks the site of the momentous
Civil War battle. There I am, at the same place where it all
happened. But wait, am I really in the same place? An observer
whose frame of reference was fixed to another planet would say
that our planet had moved through space since 1863.




Consider the figure on the facing page. The top diagram

shows the inertial motion of the earth as youd learn to represent
it if Aristotle was your schoolteacher. He tells you to use lined
notebook paper, because its blue parallel lines represent a state
of rest. The series of positions of the earth has to be drawn so
that it lines up correctly. The earth is, after all, at rest.
But then your parents pull you out of the strict, structured
Aristotelian Academy and enroll you in the hippie-dippy, coloroutside-the-lines Galilean School for Groovy Boys and Girls. At
the Galilean School, the teachers just hand out blank pieces of
white typing paper. Your new teachers tell you its all right
to make your diagram however you like, as long as the earths
positions form a line.
In other words, the things weve been saying about motion
can also be taken as claims about the structure of the background, the stage of space and time on which motion is played
out. We call this background spacetime, and each point on it
is an event: a certain time and place, such as July 1, 1863 at
Gettysburg, Pennsylvania.
Aristotelian spacetime comes equipped with a special structure, represented by the blue lines on the notebook paper. They
represent a preferred frame of reference, and they always allow
us to decide unambiguously whether or not two events happened
at the same place. Galileo corrected this view by showing that
spacetime didnt actually have any such structure. We cant say
in any absolute sense whether two events happen at the same
place. Its a matter of opinion, because it depends on the frame
of reference we pick.
But both Galileo and Aristotle implicitly assumed a further
type of structure for spacetime, which we can visualize as a set
of horizontal lines on the paper. Events lying on the same horizontal line are simultaneous, and the assumption has been that
simultaneity is not a matter of opinion. Its absolute.





Neither Galileo nor Aristotle realized that this was a nontrivial assumption, but Isaac Newton (1642-1726) did, and he
explicitly stated it:
Absolute, true, and mathematical time, of itself, and
from its own nature, flows uniformly without regard
to anything external [...]

The photo shows a 1971 experiment that proves this hidden assumption to be wrong. There had been plenty of earlier evidence, but this experiment is particularly direct and easy
to explain. Physicist J.C. Hafele and astronomer R.E. Keating brought atomic clocks aboard commercial airliners and flew
around the world, once from east to west and once from west to
east. The clocks took up two seats, and two tickets were bought
for them under the name of Mr. Clock.
When they got back, Hafele and Keating observed that there
was a discrepancy between the times measured by the traveling



clocks and the times measured by similar clocks that stayed home
at the U.S. Naval Observatory in Washington. The effect was
small enough that it has to be expressed in units of nanoseconds
(ns), or billionths of a second. The east-going clock lost time,
ending up off by 59 ns, while the west-going one gained 273 ns.
Although the effects were small, they were statistically significant
compared to the clocks margins of error of about 10 ns.
Actually it makes sense that the effects were small. Galileos
description of spacetime had already been thoroughly tested by
experiments under a wide variety of conditions, so even if its
wrong, and were going to replace it with some new theory, the
new theory must agree with Galileos to a good approximation,
within the Galilean theorys realm of applicability. This requirement of backward-compatibility is known as the correspondence
Its also reassuring that the effects on time were small compared to the three-day lengths of the plane trips. There was
therefore no opportunity for paradoxical scenarios such as one in
which the east-going experimenter arrived back in Washington
before he left and then convinced himself not to take the trip. A
theory that maintains this kind of orderly relationship between
cause and effect is said to satisfy causality.
Hafele and Keating were not working in the dark. They already had access to an improved theory of spacetime, which was
Albert Einsteins theory of relativity, developed around 19051916. They were testing relativitys predictions (and they were
not the first to do so). Lets work backward instead, and inspect the empirical results for clues as to how time works. The
east-going clock lost time, while the west-going one gained. Since
two traveling clocks experienced effects in opposite directions, we
can tell that the rate at which time flows depends on the motion
of the observer. The east-going clock was moving in the same
direction as the earths rotation, so its velocity relative to the
earths center was greater than that of the clock that remained
in Washington, while the west-going clocks velocity correspondingly reduced. The fact that the east-going clock fell behind, and



the west-going one got ahead, shows that the effect of motion is
to make time go more slowly. This effect of motion on time was
predicted by Einstein in his original 1905 paper on relativity,
written when he was 26.

If this had been the only effect in the Hafele-Keating experiment, then we would have expected to see effects on the eastgoing and west-going clocks that were equal in size. In fact, the
two effects are unequal. This implies that there is a second effect
involved, simply due to the planes being up in the air. This was
verified more directly in a 1978 experiment by Iijima and Fujiwara, in which one clock was kept in Mitaka, a suburb of Tokyo,
while an identical clock was driven up nearby Mount Norikura,
left motionless there for about a week, and then brought back
down and compared with the one at the bottom of the mountain. Their results are shown in the graph. This experiment,
unlike the Hafele-Keating one, isolates one effect on time, the
gravitational one: times rate of flow increases with height in a
gravitational field. Einstein didnt figure out how to incorporate gravity into relativity until 1915, after much frustration and
many false starts. The simpler version of the theory without



gravity is known as special relativity, the full version as general

relativity. Well restrict ourselves to special relativity for now,
and that means that what we want to focus on right now is the
distortion of time due to motion, not gravity.
By the way, the effects described in these atomic clock experiments could have seemed obscure to laypeople in the 1970s,
but today they are part of everyday life, because the GPS system
depends crucially on them. A GPS satellite in orbit experiences
effects due to motion and gravity that are both much larger than
the corresponding effects in the Hafele-Keating and Iijima experiments. The satellites carry atomic clocks, and beam down
time-stamped radio signals to receivers such as the ones used by
motorists and hikers. By comparing the time stamps of signals
from several different satellites, the receiver can calculate how
long the signals took to travel to it at the speed of light, and
therefore determine its own position. The GPS network started
out as a US military system, and this was why Hafele and Keating were able to get funding from the Navy. There is a legend
that the military brass in charge of the program werent quite
sure they believed in all the crazy relativity stuff being spouted
by the longhaired physics professors, so they demanded that the
satellites have special software switches designed into them so
that the relativity corrections could be turned off if necessary. In
fact, both special and general relativity are crucially important
to GPS, and the system would be completely broken without
We can now see in more detail how to apply the correspondence principle. The behavior of the clocks in the Hafele-Keating
experiment shows that the amount of time distortion increases as
the speed of the clocks motion increases. Galileo lived in an era
when the fastest mode of transportation was a galloping horse,
and the best pendulum clocks would accumulate errors of perhaps a minute over the course of several days. A horse is much
slower than a jet plane, so the distortion of time would have
had a relative size of only one part in 1, 000, 000, 000, 000, 000
(1015 ) much smaller than the clocks were capable of detect-



ing. At the speed of a passenger jet, the effect is about one part
in 1,000,000,000,000 (1012 ), and state-of-the-art atomic clocks in
1971 were capable of measuring that. A GPS satellite travels
much faster than a jet airplane, and the effect on the satellite
turns out to be about one part in 10 billion (1010 ). The general idea here is that all physical laws are approximations, and
approximations arent simply right or wrong in different situations. Approximations are better or worse in different situations,
and the question is whether a particular approximation is good
enough in a given situation to serve a particular purpose. The
faster the motion, the worse the Galilean approximation of absolute time. Whether the approximation is good enough depends
on what youre trying to accomplish. The correspondence principle says that the approximation must have been good enough to
explain all the experiments done in the centuries before Einstein
came up with relativity.
But I dont want to give the impression that relativistic effects are always small. If we want to see a large time dilation
effect, we cant do it with something the size of the atomic clocks
Hafele and Keating used; they would become deadly missiles with
destructive power greater than the total megatonnage of all the
worlds nuclear arsenals. We can, however, accelerate subatomic
particles to very high speeds. For experimental particle physicists, relativity is something you do all day before heading home



and stopping off at the store for milk. An early, low-precision experiment of this kind was performed by Rossi and Hall in 1941,
using naturally occurring cosmic rays. The figure shows a 1974
particle-accelerator experiment of a similar type.

Particles called muons (named after Greek , myoo) were

produced by an accelerator at CERN, near Geneva. A muon
is essentially a heavier version of the electron. Muons undergo
radioactive decay, lasting an average of only 2.197 microseconds
(millionths of a second, coincidentally also written with the Greek
). The experiment was actually built in order to measure the
magnetic properties of muons, but it produced a high-precision
test of time dilation as a byproduct. Muons were injected into
the ring shown in the photo, circling around it until they underwent radioactive decay. At the speed at which these muons were
traveling, time was slowed down by a factor of 29.33, so on the average they lasted 29.33 times longer than the normal lifetime. In
other words, they were like tiny alarm clocks that self-destructed
at a randomly selected time. The graph shows the number of
radioactive decays counted, plotted versus the time elapsed after a given stream of muons was injected into the storage ring.
The two dashed lines show the rates of decay predicted with and
without relativity. The relativistic line is the one that agrees
with experiment.




The principle of greatest time

A paradox now arises: how can all of this be reconciled with the
fact that motion is relative? If clocks run more slowly because
theyre in motion, couldnt we determine a preferred rest frame
by looking for the frame in which clocks run the fastest?
To clear this up, it will help if we start with a simpler example
than the Hafele-Keating experiment, which involved three different sets of clocks, motion in circles, and the motion of the planes
being superimposed on the rotation of the earth. Lets consider
instead the relativistic version of the Galilean twin paradox from
section 1.6, p. 21, reproduced here in panel a of the figure. Recall that in this thought experiment, Alice stays at home on earth
while her twin Betty goes on a space voyage and returns. The
relativistic version of this story is in fact what people normally



have in mind when they refer to the twin paradox, and it has
an especially strange twist to it. According to relativity, time
flows more slowly for Betty, the traveling twin. The faster her
motion is, the bigger the effect. If she flies fast enough, we can
produce a situation in which Betty arrives home still a teenager,
while Alice has aged into an old woman!
But how can this be? In Bettys frame of reference, isnt it the
earth thats moving, so that it should be the other way around,
with Alice aging less and Betty more?
The resolution of the paradox is the same as in the Galilean
case. Alices motion is inertial, and we can see this because her
track across the diagram, called her world-line in relativistic parlance,1 is a straight line. Bettys motion isnt inertial; it consists
of two segments at an angle to one another. In fact, there is a
nice way of stating the relativistic rule for time dilation in these
terms. The principle of greatest time states that among all the
possible world-lines that a clock can follow from one event to another, the one that gives the greatest elapsed time is the inertial
one. The shortened time interval is said to be dilated.
Analyzing the Hafele-Keating experiment becomes almost as
easy if we simplify by pretending that the west-flying plane was
going at exactly the right speed to cancel out the earths rotation.
(In reality this was probably roughly, but not quite, true.) Then
this plane actually stood still relative to the center of the earth,
whose motion is inertial. Therefore the planes motion is inertial,
as shown in panel b of the figure. The east-going clock flies in
a circle, with its airspeed added on to the rotation of the earth
to give double the velocity. As shown in panel c, its world-line is
not an inertial straight line. Although panels b and c are drawn
separately for clarity, the clocks really did start out together at
the beginning and have a reunion at the end, as in the principle
of greatest time. Therefore the inertial, west-going will record
more elapsed time than the east-flying plane.

The odd-sounding term is a translation from German of a phrase intended to mean line through the world, i.e., through spacetime.




A universal speed limit

So far weve considered experiments that seem to fall into two

different categories. In the atomic clock experiments, the clocks
were together to start with, were separated, and were then reunited and compared again. These dovetail nicely with the structure of the principle of greatest time. But no such reunion happens, for example, in the muon experiments. When I first learned
relativity, I got terribly confused about the no-reunion experiments. According to Einstein, if observers A and B arent at rest
relative to each other, then A says Bs time is slow, but B says A
is the slow one. How can this be? If A says B is slow, shouldnt
B say A is fast? After all, if I took a pill that sped up my brain,
everyone else would seem slow to me, and I would seem fast to
Suppose, for example, that Betty is on the outward-bound leg
of her interstellar taco run. She puts the ship on cruise control
and gets on the phone with Alice. Talking on the phone, cant
they now establish whos actually slow and whos fast? Shouldnt
the slow one sound like Darth Vader to the fast one, and conversely shouldnt the slow one hear the fast ones voice as that of
a chipmunk whos been using helium and crack cocaine? There
would then be an objective difference between Alice and Bettys
frames but this wrecks the logic of relativity, which says that
all inertial frames are supposed to be equally valid.
The key is that weve been implicitly assuming that the interstellar cell phone is an instantaneous method of communication,
so that it establishes the truth of the matter: who is really slow
and who is really fast. We are forced to the opposite conclusion,
that the phones do not send signals that propagate at an infinite
speed in fact, that no form of communication, no method of
cause and effect, can operate at a distance without a certain time
lag. The whole logical structure of relativity falls apart unless we
assume a universal speed limit for all motion in the universe. We
refer to this speed with the letter c.
The figure shows what actually happens to Alice and Betty



if, for example, one twin makes two hand claps near her phone,
separated by a one-second interval. If the two signals traveled
at infinite speed, then their world-lines would be horizontal
they would be received at the same time they were sent. But
because they actually travel at a finite speed (lets say exactly
at c), they are sent at one time and arrive at some later time,
and the signals world-lines have some slope. We conventionally
choose the time and distance scales on our diagrams so that this
is a 45-degree angle. We can see on the diagram that when Alice
sends the two hand claps, Betty receives them at times that are
spread to more than one second apart, but exactly the same thing
happens when we flip the diagram. The situation is completely
symmetric, and each twin perceives the others transmission as
having been slowed down.
The universal speed limit c is also the speed at which light
travels. In the metric system, which is designed to handle times
and distances on the human scale, c is a huge number, about
300, 000, 000 meters per second (3 108 m/s in scientific notation). This is an example of the correspondence principle at work.
One of the reasons that we dont notice the effects of Einsteins
relativity (relativistic effects) very often in everyday life is that
the speeds at which we walk, drive a car, and throw baseballs are
so small compared to c.




Einsteins train

Weve seen that if simultaneity isnt absolute, then there must

be a universal speed limit c. The converse is also true: if c is
universal, then simultaneity must be relative. The figure shows
a famous thought experiment used by Einstein to present this

A train is moving at constant velocity to the right when bolts

of lightning strike the ground near its front and back. Alice,
standing on the dirt at the midpoint of the flashes, observes that
the light from the two flashes arrives simultaneously, so she says
the two strikes must have occurred simultaneously.
Bob, meanwhile, is sitting aboard the train, at its middle. He
passes by Alice at the moment when Alice later figures out that
the flashes happened. Later, he receives flash 2, and then flash
1. Since the light from the flashes travels at c, and c is universal,
it has the same value in Bobs frame of reference as in Alices.
Bob knows that the two flashes traveled at the same speed as
each other. Therefore he infers that since both flashes traveled
half the length of the train, and flash 2 arrived first, flash 2 must
have occurred first. The two events that in Alices frame are
simultaneous are not simultaneous to Bob.


Velocities dont simply add

Its crucial to the example of the train in section 2.4 that the
speed of light is the same for Bob as for Alice. We saw in section
1.5, p. 19, that according to the Galilean description of spacetime,



velocities add in relative motion. This is not the case according

to Einsteins relativity. If it were, then a velocity c in one frame
wouldnt equal c in another frame. It wouldnt be universal.
Einstein later recalled that when he was trying to create special
relativity, he got stuck on this point for about a year. It stumped
him that the concept of the invariance of the velocity of light
. . . contradicts the addition rule of velocities, which he had been
assuming would still be true. Finally he realized that Time
cannot be absolutely defined, and there is an inseparable relation
between time and [...] velocity. After this he finished his seminal
paper on special relativity within five weeks.
Addition of velocities had, of course, already been used successfully in countless experiments and practical applications in
the course of the several centuries since Galileo. The correspondence principle tells us that simple velocity addition must be a
good approximation in the conditions under which it had already
been tested. But all of these experiments had been ones involving
material objects at speeds low compared to c. Well see later how
to combine velocities correctly, but that method will have to give
results nearly the same as straight addition when the velocities
are small. For example, in section 1.5 we added velocities of 30
km/hr and 20 km/hr to get 50 km/hr. The correct relativistic
result is 49.9999999999999995 km/hr.
That was for velocities small compared to c. If we go to the
opposite extreme, combining a velocity that equals c with any
other velocity should simply give back c, since c is universal. A
high-precision test of this prediction was performed in 1964 by
Alvager et al. In their experiment, subatomic particles called
mesons were produced, in motion at 99.98% of c. These particles
are radioactive and decay in flight, producing flashes of light. If
Galilean addition of velocities were correct, then a flash emitted
in the forward direction at 100% of c relative to the decaying
particle would be expected to move with a velocity of 199.98% of
c, i.e., almost twice the normal speed of light. The experiment
actually found them to travel at 100.00% of c, as predicted by
relativity, to within the experiments precision of 0.01%.




The light cone

In figure a above, a light bulb blinks on momentarily and then

back off. Flashes of light spread out from it at c to the right
and left, tracing lines which, because of our choice of units, make
45-degree angles. These lines form a geometrical figure, which
is called the light cone, for reasons that may be more clear from
figure b, which attempts to represent two dimensions of space as
well as the time dimension.
You might not guess from this description that the light cone
is fundamentally important, or that it doesnt have much to do
with light. Remember, c isnt really the speed of light, its the
ultimate speed limit on cause and effect. We can draw a light
cone centered on any event, such as the one labeled ? in the
figure on the next page. If ? is the Russian Revolution, then
events inside ?s future light cone can be described as all the
ones that we could reach, starting from the Russian Revolution,
by traveling at less than the speed of light. A signal could be sent
from ? to any of these events, and ? could be the cause of any
of these events. In Einsteins theory of space and time, the light
cone plays the same fundamental geometrical role as the circle in
Euclidean geometry.
Events inside ?s light cone are said to have a timelike relation
to ?, because they are separated from it by more time than space.
For example, one of these events could be separated in time from



? by 100 years, but in space by only 50 light-years (the distance

light travels in 50 years). Then a spaceship could get there by
going at half the speed of light. Timelike relationships can be
future-timelike or past-timelike. Events outside ?s light cone
are said to be spacelike in relation to ?, and events on the surface
of the light cone as lightlike.





Scientific revolutions

Special relativity overthrew Galilean relativity, a scientific theory

that physicists had wholeheartedly believed in for hundreds of
years. Does an example like this mean that physicists dont know
what theyre talking about?
Some social scientists deny that the scientific method even
exists, claiming that science is no more than a social system
that determines what ideas to accept based on an in-groups criteria. If so, it would seem difficult to explain its effectiveness
in building such useful items as CD players and sewers accomplishments that seem to have eluded voodoo and astrology.
The extreme social-relativist attitude was effectively skewered
in a famous hoax carried out in 1996 by New York University
physicist Alan Sokal, who wrote a nonsense article titled Transgressing the Boundaries: Toward a Transformative Hermeneutics
of Quantum Gravity, and got it accepted by a cultural studies
journal called Social Text.
Most physicists claim that their science does tell us objective
truths about the world, and they would invoke the correspondence principle (p. 31) in order to explain an example such as the
replacement of Galilean relativity with special relativity. From
this point of view, science is not merely a matter of fashion. If a
theory such as Galilean relativity became accepted in physics, it
was because it was supported by a vast number of experiments.
Its just that experiments never have perfect accuracy. The old
experiments werent all wrong. They were right, within their
limitations. This point of view is associated with the philosopher
of science Karl Popper (1902-1994). Trained as a psychologist
during the era when Freud was a dominant figure, Popper proposed falsifiability as the criterion for distinguishing science from
non-science. According to Popper, real science makes itself vul-



nerable by proposing theories that predict new things, and that

could be disproved if the predictions turned out to be false. He
saw Freudianism as nonscientific, since it was so elastic that it
could be made to retroactively explain any facts whatsoever.
It can be difficult to shoehorn real science into Poppers system. Galileo got in trouble for saying that the solar system revolved around the sun rather than the earth, and in order to
advocate this, he had to argue that motion was relative, since
otherwise we would observe dramatic effects of the earths motion in our environment at the earths surface. But in fact the
sun-centered Copernican system did not make more accurate predictions of the motion of the planets than the earth-centered one.
By Poppers criterion, the whole argument was unscientific.

A controversial alternative to Popper comes from Thomas

Kuhn (1922-1996), who argues that science operates in two different modes, a normal one in which knowledge is steadily accumulated, and revolutionary steps, or paradigm shifts. According
to Kuhn, Galilean relativity and heliocentrism were a paradigm
shift that only much later led to a theory that was superior to the
old one in terms of falsifiability. Kuhn claims that theories from
after a paradigm shift are incommensurable with theories from
before it, meaning that they ask different questions and require
different world-views of their users. He compared this to an optical illusion that the brain can interpret in two different ways.
Kuhn would probably reject my analysis in terms of the correspondence principle as a fairy-tale version of history. His critics
see him as attacking scientific realism, the notion of scientific
truth, and the authority of science.



Chapter 3

The Lorentz

Surveying space and time

The human brain is highly specialized for visual processing, and

we have the intuitive feeling that we can always look at a scene
and get a snapshot of what is, a continuously updated flicker of
now on our retinas. Metaphorically, we even apply this intuition
to things that arent actually visible. I see what you mean, we
say. Lets look at the situation. Are we seeing eye to eye?
Just as Galileos contemporaries struggled to overcome their
intuitive resistance to the principle of inertia, people today need
to wrestle with this powerful intuition, which is wrong and arises
from the fact that the light coming to our eyes goes so fast that
it seems to arrive instantaneously. When you look at the night
sky, youre seeing the stars as they were in the past, when their
light started on its journey to you. A star may have already gone
through its death throes and exploded, but still appear to us as
it was before. Within an atom, the subatomic particles move
at a substantial fraction of c, and attract and repel each other
through electric and magnetic forces. The force experienced by
a particular electron is not a force created by the other particles



at their current locations, but at the locations they had before

the time lag required for the propagation of the force. In the
movie Star Wars, a planet is destroyed by the bad guys, and Obi
Wan Kenobi says, I felt a great disturbance in the Force, as if
millions of voices suddenly cried out in terror and were suddenly
silenced. But if the Force operated according to the same
rules as the known fields of force in physics (electricity, gravity,
etc.), there would have been a delay of many years before he
could have gotten the signal.
Once we do receive a signal from some distant event, we can
always apply a correction to find out when it really happened.
For example, Chinese astronomers recorded a supernova in the
year 1054, in the constellation Taurus. We now know that the
star that exploded was about 6000 light years from earth. (A
light year is the distance light travels in one year.) Therefore
the explosion must have really happened around the year 5000
B.C. The now wasnt really now, it was a long time ago. This
correction technique is called Einstein synchronization. But even
with Einstein synchronization, the example of the train, section
2.4, p. 40, shows that simultaneity is relative. If observers A
and B are not at rest relative to one another, they have different
definitions of now.


The Lorentz transformation

Subject to these limitations, the best we can do is to construct

the relativistic version of the Galilean transformation (section
1.7, p. 22). The relativistic one, shown in the figure on the
next page, is called the Lorentz transformation, after Hendrik
Lorentz (1853-1928), who partially anticipated Einsteins work,
without arriving at the correct interpretation. The distortion of
the graph-paper grid is a kind of smooshing and stretching.
The most noticeable feature of the Lorentz transformation
is that it doesnt maintain simultaneity. The horizontal lines on
the white grid connect points that are simultaneous in the earths



frame of reference, but they are not parallel to the corresponding

lines on the black grid, which represents the rocket ships frame.
One thing that does coincide between the two grids is the 45degree diagonal, which is a piece of the light cone. This makes
sense, because the two observers should agree on c. As usual,
the units of the graph are chosen so that c lies at this slope. For
example, if the units of time are years, then the distances have
to be in light-years.
As in the Galilean case, the slope of the black grids left edge
relates to the velocity of the spaceship relative to the earth. In
this example, the spaceship takes 5 time units to travel 3 distance
units, so its velocity relative to the earth is 3/5 of c.




Correspondence principle

The correspondence principle requires that the Lorentz transformation (p. 49) match up with the Galilean transformation
(p. 22) when the velocity of one frame relative to the other is
small compared to c. This does work out properly because of
the special relativistic units used in our graphical representation
of the Lorentz transformation. Lets say that were using seconds as our unit of time, so that our unit of distance has to be
the light-second, the distance traveled by light in one second.
A light-second is a huge distance, roughly equal to the distance
from the earth to the moon.

The figure above shows an example of what happens in a

nonrelativistic experiment. In panel a, we have a Lorentz transformation for a velocity thats small compared to c, in relativistic
units. In panel b, we stretch the horizontal scale quite a bit, although nowhere near as much as we would need to do in order to
convert from, say, light-seconds to meters. The result is nearly
indistinguishable from a Galilean transformation.


Scaling, and the squish and stretch factors

The physical arguments weve made so far are enough to pin

down every feature of the Lorentz transformation except for an
over-all scaling factor. Given the white grid, we could seemingly
take the black one and blow it up or reduce it like a photograph.



This scale can be fixed by using the fact, proved in section 3.11,
p. 61, that areas have to stay the same when we carry out a
Lorentz transformation.
This makes it possible to give a very simple geometrical description of the Lorentz transformation. Its done by stretching
one diagonal by a certain amount and squishing the other diagonal by the inverse of that factor, so that the total area stays the
same. The figure shows the case where the velocity is 3/5 of c,
which happens to give a stretch by a factor of 2 and a squish by


No acceleration past c

The stretch and squish factors give us an easy way to visualize

what happens if we try to keep accelerating an object past the
speed of light. Since the numbers are simple for a velocity of 3/5,
lets imagine that we take an object and accelerate it to 3/5 of c
relative to the earth. The corresponding Lorentz transformation
is stretched and squished by a factor of 2. Now lets say that
we continue accelerating it until its moving at an additional 3/5
of c relative to that motion. If velocities added the way Galileo
imagined, then this would give 6/5 of c, which is greater than c.
But what will actually happen here is that well have stretched
the long diagonal by a factor of 4 and shrunk the short one by
the same factor (left panel of the figure below). This is a Lorentz
transformation that still corresponds to some velocity less than
c. (It turns out to be 15/17 of c.) No matter how many times we



do this, the velocity will still be less than c. Thus no continuous

process can accelerate an object past the speed of light, although
this argument doesnt rule out a discontinuous process such as
Star Treks transporter.
By about 1930, particle accelerators had progressed to the
point at which relativistic effects were routinely taken into account. In 1964, W. Bertozzi at MIT did a special-purpose experiment as an educational demonstration to test relativity using
an electron accelerator. The accelerator was powerful enough to
have accelerated the electrons to many times the speed of light,
had Galileo been right. As shown in the right panel of the figure,
the actual speeds measured just got closer and closer to c. The
experimental data-points are in good agreement with the dashed
line, which shows the predictions of special relativity.


Time dilation

The figure shows a close-up of a Lorentz-transformation graph,

covering just a little more than one unit of time one hour.
The white and black clocks are in motion relative to one another.
If the observer in the white frame sends signals back and forth
and goes through the process of Einstein synchronization (dashed
line), she finds out that at the moment when the black clock



shows one hour of elapsed time, her own clock has measured 1.25
hours. (This is again for our favorite numerical example where
the velocity is 3/5 of c.) This ratio of 1.25, or 5/4, is the factor
by which each of the two observers says that the other ones time
has slowed down. There is a mathematical symbol for this ratio,
and a formula that would allow us to plug in the 3/5 and get
the 5/4 out, but the focus of this book isnt on that kind of
numerical calculation. For our purposes, drawing the graph is
just as reasonable a way of working out the result.
In this example, the velocity was moderate compared to c,
and the time-dilation factor was moderately big. For velocities
small compared to c, the correspondence principle tells us that
time-dilation factor must get close to 1, i.e., the effect goes away.
As the velocity gets closer and closer to c, the time-dilation factor
blows up to infinity. This is the scientific background for science
fiction stories such as the Planet of the Apes movies, in which
thousands of years pass back home while astronauts aboard a
spaceship experience only a few months. This kind of motion
very close to the speed of light is referred to as ultrarelativistic.
In theory, if we had enough energy, we could accelerate ourselves
to very close to the speed of light and watch the stars grow cold
and the cosmos slip into senescence.




Length contraction

A science fiction story with better dramatic possibilities might be

one in which our protagonists zoom across the galaxy and have
various adventures, all within a human lifetime. Our galaxy is
about 10,000 light-years in size, so since the ship cant go faster
than c, the voyage would take at least 10,000 years according
to people back on earth. But due to time dilation, this time
interval would seem shorter to the travelers. If they moved at ultrarelativistic speeds, the time dilation could be extreme enough
so that they could live long enough to complete the trip within
their lifetimes.
But what does this all look like according to the voyagers?
To them, their own time seems normal, and its the people back
on earth who are slowed down. So how can they explain the
fact that they traveled 10,000 light-years such a short time? The
answer is that the distance doesnt seem like 10,000 light-years
to them. The Lorentz transformation describes distortions of
both time and space. To the astronauts, the galaxy appears
foreshortened, as suggested by the figure. This shortening of
distances, called length contraction, is by the same ratio as the
time dilation factor. Summarizing these results, we have the



Time dilation
A clock appears to run fastest to an observer at rest relative to the clock.
Length contraction
A meter stick appears longest to an observer who is at rest
relative to it.


Not what you see

By the way, you shouldnt get the impression from words like
appears that these are the only effects that one actually sees
with optical observations using the eye or a camera. What we
mean here is what an observer finds out from a procedure such as
Einstein synchronization, which involves sending signals back and
forth, completing a surveying process, and working backward to
eliminate the delays due to the time the signals took to propagate.
In an actual optical measurement, these delays are present, and
they have an effect on what we see.
Furthermore, the rays of light that come to us suffer an effect called relativistic aberration, which makes them seem like
theyre coming from a different direction than they really are.
To understand this effect, imagine, as in panel a of the figure
on the following page, that youre riding in a convertible with
the top down, at 40 km/hr. Meanwhile, rain is falling straight
down at 40 km/hr, as measured in the frame of reference of the
sidewalk. But in your frame, the rain appears to come down at
a 45-degree angle, not from straight overhead. Relativistic aberration is usually a small effect because we dont move eyes or
cameras at velocities that are comparable to the speed of light.
However, the effect does exist, and astronomers routinely take
it into account when they aim telescopes at high magnification,
because the telescope is being carried along by the motion of the
earths surface.





Panels b-d of the same figure show a visualization for an observer flying through a cube at 99% of the speed of light. In b,
the cube is shown in its own rest frame, and the observer has
already passed through. The dashed line is a ray of light that
travels from point P to the observer, and in this frame it appears
as though the ray, arriving from an angle behind the observers
head, would not make it into her eye. But in the observers frame,
c, the ray is at a forward angle, so it actually does fall within her
field of view. The cube is length-contracted by a factor about 7.
The ray was emitted earlier, when the cube was out in front of
the observer, at the position shown by the dashed outline.
The image seen by the observer is shown in panel d. Note that
the relativistic length contraction is not at all what an observer
sees optically. The optical observation is influenced by length
contraction, but also by aberration and by the time it takes for
light to propagate to the observer. The time of propagation is
different for different parts of the cube, so in the observers frame,
c, rays from different points had to be emitted when the cube was
at different points in its motion, if those rays were to reach the



Although time dilation is a kind of time travel, its always time

travel into the future, so special relativity avoids scenarios such
as the Robert Heinlein story All You Zombies , in which
a woman has a sex-change operation, goes back in time, has
sex with a woman who he doesnt realize is his younger self, and
becomes the father of a baby who turns out to be himself. If event
A causes a later event B in a certain frame of reference, then B
must be inside As future light cone. The Lorentz transformation
doesnt change the light cone, so B is guaranteed to be later than
A in all other frames of reference as well.




The stretch factor is the Doppler shift

If you hear little kids playing with toy race cars, theyll often
imitate the sound theyve heard on TV as the cars go by: the buzz
of the engine starts out a little higher in pitch as it approaches,
then shifts a little lower as it goes away. This is called a Doppler
shift. The size of the effect depends on how fast the car is moving
compared to the speed of sound. We dont notice Doppler shifts
as much for ordinary cars, because their speeds are much smaller
compared to the speed of sound. The revolutions of the spinning
motor are like the ticks of a clock. The rate at which these sound
impulses arrive at the ear of a stationary observer is distorted
because between the emission of one impulse and the next, the
distance to the ear has changed.
We can also observe Doppler shifts with light rather than
sound. The concept turns out to be very closely related to the
situation illustrated in the figure on p. 59. This was a spacetime
diagram for a situation in which Alice and Betty, flying away from
each other in spaceships, try and fail to establish whose time is
really slow and whose fast. In that experiment, Alice clapped her
hands twice, one second apart, next to her cell phone, and Betty
heard the hand claps more than one second apart. Although
Betty might conclude from this that Alices time is slow, the
situation is symmetric, and Alice gets the same results if Betty
sends hand-claps to her.
The figure on the facing page shows what happens if we repeat
the hand claps over and over. The claps begin while Alice and
Betty are approaching one another, and continue after they do
their flyby and recede. While theyre approaching, the hand claps
are received closer together in time, and this can be interpreted
as a combination of two effects: their earlier and earlier arrival
as the distance shrinks, and a time dilation effect, which only
partially counteracts the distance effect. As they recede, the
distance effect causes the time intervals to be wider, and the
time dilation effect cooperates with the diatance effect to make
the intervals even longer.



But now suppose that the signal is not a series of claps but
simply a radio wave. We previously thought of the diagonal
world-lines as the hand-clap signals traveling through space. Now
we can think of them as the peaks of the wave traveling through
space. During the receding part of the motion, the peaks are
farther apart in time when received, so the receiver perceives the
wave as having slower vibrations (a lower frequency). This is
similar to the Doppler shift of the race cars sound, but what
matters now is how fast the ships are traveling compared to the
speed of light.
The mathematical description of this type of Doppler shift is
very simple. The motion of A relative to B is at some speed, and
at this speed the Lorentz transformation has a certain stretch
and squish factor (section 3.4, p. 50). Suppose that the stretch
and squish factors are 2 and 1/2. Then the Doppler shift changes
the rate at which the signals are received by a factor of 2 if A and



B are approaching one another, or by 1/2 if theyre receding.

Because the velocities in everyday life are so small compared
to the speed of light, the Doppler shifts are very small. For
example, if a jet plane is flying directly toward us at 300 meters
per second, the light that comes from it to our eyes is shifted up in
frequency by a factor of 1.000001, which is only a ten thousandth
of a percent, so although the light is shifted a little bit toward
the blue end of the spectrum, the change isnt noticeable. As
it recedes from us, the frequencies get multiplied by 0.999999,
which is a drop of a ten thousandth of a percent, and we have a
tiny shift toward the red.
But there is at least one example in everyday life where
Doppler shifts matter. When you use GPS, your receiver is getting signals from satellites that are moving about ten times faster
than a jet plane. Although this is still very small compared to the
speed of light, GPS depends on extremely accurate time measurements, so the Doppler shifts are big enough so that if we didnt
take them into account, the system wouldnt work at all.
Because the Lorentz transformation preserves the area of a
square as it distorts it into a parallelogram, the stretch and shift
factors are always inverses of one another: 2 and 1/2, 7 and 1/7,
and so on. This exact relationship would not be true without
relativity, so testing it provides a test of relativity. Physicists using specialized equipment can measure Doppler shifts extremely
accurately, so these tests are among the strictest to which anyone
has ever subjected the theory. The first such test was done by
Ives and Stilwell in 1938, and the results were as predicted by
relativity. A modern, more precise version by Saathoff et al. in
2003 used lithium ions moving at 6.4% of c in a particle accelerator. The forward and backward Doppler shifts were measured to
be 1.06592034 and 0.93815641, respectively, and if you multiply
these two numbers together, youll see that the product equals 1
to nine digits of precision, which is impressive by anyones standards. The spectacular agreement with theory has made this
experiment a lightning rod for anti-relativity kooks.
As well see in chapter 13, p. 165, Doppler shifts are the


method we use to measure the motion of distant galaxies toward or away from us, and this was the technique by which we
discovered that the universe was expanding.


Proof that area stays the same in a

Lorentz transformation

This optional section gives a proof that Lorentz transformations

dont change area in the t x plane
We first subject the square in the figure to a transformation
with velocity v, and this increases its area by a factor F , which
we want to prove equals 1. We chop the resulting parallelogram
up into little squares and finally apply a v transformation, i.e.,
a Lorentz transformation with the same speed but in the opposite



direction. This changes each little squares area by a factor G, so

that the whole figures area is also scaled by G. The final result
is to restore the square to its original shape and area, so F G = 1.
But we must have F = G, because space is symmetrical, and we
cant treat one direction differently than theother. Since F and
G are both positive, they must both equal 1, which was what we
wanted to prove.

Chapter 4

The measure of all



In college, I was a geeky teenager who was high on math and

physics. My father and stepmother lived close enough that I
could stop by for dinner on the weekend, and often when I showed
up my stepmother would sit me down in their tiny kitchen, pour
glasses of white wine for me and herself, and engage me in conversation. Having majored in English, with a lifelong interest in
ballet, she must have decided to try to at least partially ungeekify me. On one of these occasions, I announced to her, trying
to be interesting and provocative, that nothing really existed unless you could measure it. What about love? she asked. I
cant remember whether I replied that love could be measured or
that love didnt exist probably the latter, since it would have
sounded more provocative.
Despite my immature understanding of it, operationalism, developed by Bridgman around 1927, is a legitimate approach to
understanding what we mean by the words we use to describe
the world around us. The operationalist approach works like
this. What is time? Time is what a clock measures. That is,



measuring time requires carrying out certain measurement operations, and, according to Bridgman, we mean by any concept
nothing more than a set of operations; the concept is synonymous
with the corresponding set of operations. It may not work for
love, but it does work pretty well for time.


Galilean scalars and vectors

Most of the things we want to measure in physics fall into two

categories, called vectors and scalars. In Galilean relativity, a
scalar is something that doesnt change when you turn it around,
while a vector does change when you rotate it, and the way in
which it changes is the same as the way in which a pointer such as
a pencil or an arrow would change. For example, temperature is a
scalar, because a hot cup of coffee doesnt change its temperature
when we turn it around. Force is a vector. When I play tug-ofwar with my dog, her force and mine are the same in strength,
but theyre in opposite directions. If we swap positions, our forces
reverse their directions, just as an arrow would.

The simplest example of a Galilean vector is a motion from

one place to another, called a displacement vector.
I chose the coffee and tug-of-war examples because they were
intuitively appealing, but we dont have to depend on intuition
to check that theyre right. For example, temperature can be
defined operationally as what a thermometer measures, so one
can actually put a thermometer in a cup of coffee, rotate the
cup, and check that the thermometer reading stays the same.



Scalars are just numbers, and we do arithmetic on them in

the usual way. With vectors, the rules are a little different. In
particular, there is a special method for adding vectors, as shown
in the figure below. We represent them with two arrows, slide
the arrows so that the tip of one is at the tail of the other, and
then draw the result as the arrow that goes from the tail of the
first vector to the tip of the second one. If we hike from A to B,
and then from B to C, vector addition shows us how we could
have achieved the same result by hiking in a straight line from A
to C, without the detour to B. When we add two force vectors,
we get a single force vector that would give the same result, if
it acted on a certain object, as the two individual vectors would
have given if they had both acted on that object.

A vector has a magnitude, which means its size, length, or

amount. Rotating a vector can change the vector, but it will
never change its magnitude. In fancy language, we say that the
magnitude of a vector is invariant under rotations. A meter-stick
is still a meter-stick, no matter how we turn it around.


Relativistic vectors and scalars

But in relativity, a meter-stick might not be a meter long. Its

length could be Lorentz contracted. This is a sign that we need
to rethink the idea of vectors and scalars a little in order to make
them useful in relativity. In relativity we think of space and time
as a single, unified stage on which events play out. The basic
example of a vector is now a displacement in spacetime. Such



a displacement could take me from here to Buenos Aires, or it

could take me from now to next week. Imagine that you have a
magic doorway that can take you to any time or place. You have
to twiddle some knobs to tell it where and when. The settings
on the knobs specify a relativistic displacement vector.
Such a vector could change not just because an observer
viewed it from a different direction, but also because the observer was in a different state of motion. The direction change is
a rotation. The change in motion is referred to as a boost. A
relativistic scalar is something that doesnt change under rotations or boosts. A good example of a relativistic scalar is electric
charge, the property that makes your socks cling together when
they come out of the dryer. Although time is a Galilean scalar,
its not a relativistic scalar, because when we do a boost, a time
interval can suffer time dilation. A time interval by itself is neither a relativistic vector nor a relativistic scalar; it could be part
of the description of a spacetime displacement.


Measuring relativistic vectors

We would like to be able to measure relativistic vectors and describe their magnitudes in the same way we did with Galilean
vectors. We can do that, because Einsteins relativity, unlike
Galileos, gives us a way of expressing times and distances in the
same system of units.
An example is the relativistic displacement vector from your
birth to your death. As you live your life, your body traces
out a world-line through spacetime. You cant go faster than c,
so for example the displacement from your birth to your death
is guaranteed to have a time component that is bigger than its
spatial part. If you live to be 80 years old, your place of death
is guaranteed to be less than 80 light-years from your place of
birth. (In the earths frame of reference, its much, much less.)
In the terminology of section 2.6, p. 42, these events are therefore
timelike in relation to each other. The vector connecting them is



likewise called a timelike vector.

The magnitude of a timelike vector is defined as the time
elapsed on a clock that moves inertially along that vector. This
clock-time is called proper time. The word proper is used here
in the somewhat archaic sense of own or self, as in The
Vatican does not lie within Italy proper. Suppose that an alien
living in a distant galaxy is observing you through a powerful
telescope. Because the universe is expanding, she sees our galaxy
as moving away from her at a significant fraction of the speed of
light. Your hypothetical lifetime of 80 years is dilated as seen
in her frame of reference, and may be 200 years to her. But
if she knows your motion, she can correct for that, and infer
that a clock moving along with you would only have registered
an elapsed time of 80 years. You and the alien agree on the
magnitude of your birth-to-death vector: its 80 years.
If we imagine the alien as moving relative to us (and us relative to her) at a speed very, very close to c, the correction factor
can get very large. It might take a trillion years of telescope
observing for the alien to see you finish brushing your teeth this
morning. If some thing, say a ray of light, moves at exactly c,
then its proper time becomes zero for any observer. In this sense,
a ray of light experiences no time at all. This type of displacement is lightlike, and its magnitude is defined to be zero, even if
the vector itself isnt zero.

Spacelike vectors are of less interest because they cant trace

out the world-line of any thing. Although it is possible to define
a magnitude for a spacelike vector, we wont go into that now
because it leads to complications involving plus and minus signs.



Part II

Matter and E = mc2


Chapter 5

Light and matter before


How much does matter really matter?

Everyone knows E = mc2 , even if they dont know what the letters stand for. You already know about c, so lets move on to
m. It stands for mass, and most dictionaries define it as something like the quantity of matter. This definition falls into the
category referred to by the physicist Wolfgang Pauli as not even
wrong, meaning that it doesnt even mean enough for us to decide whether its right. The basic problem is that its not an
operational definition. As an operational definition, pick up the
nearest handy heavy object, such as your backpack full of textbooks. Hold it firmly in both hands, and shake it back and forth.
What youre feeling is a sensation of its mass. If you were in outer
space, where there is no gravity, youd get exactly the same results, because mass doesnt depend on gravity. This is in fact a
technique sometimes used by astronauts to check on their body
mass while theyre in orbit. They sit in a chair attached to a
spring, and measure how fast the vibrations are.
As an interpretation (not a definition), we can say that mass
is a measure of how much inertia an object has: how hard it is



to change its state of motion.

To define a numerical scale of mass, we can just start with
a standard object, and do this kind of vibration measurement
on the standard object. (The metric kilogram unit, for example,
is defined by a standard platinum-iridum cylinder housed in a
shrine in Paris.) If the results are the same for the standard
object as for our object of unknown mass, then the unknown has
a mass of one unit. If three copies of the standard object bundled
together give the same results as our unknown, then its mass is
three units.


How hard is the hit, how loud is the


Mass measures how hard it would be to change an objects motion, but we would also like to have some way to measure how
much motion it actually has. We could measure its velocity,
but that would give the same measure of motion to worldchampion sprinter Usain Bolt as to my skinny little spanielgreyhound mutt, who can run at about the same pace. Bolt
is a big, burly guy, and if I had to choose whether to get hit by
him or my dog, running at full speed, Id choose the dog. What
were groping for here is something called momentum. As an
operational definition, momentum can be defined by taking the
moving object (say my dog), colliding it with a target (me), and



seeing how fast the target recoils. If stopping Usain Bolt makes
me recoil ten times faster than stopping my little dog, then Bolt
has ten times more momentum. This definition is exactly the one
used to measure momentum in particle-accelerator experiments.
At nonrelativistic speeds (i.e., speeds much lower than c), experiments show that momentum is proportional to the objects
mass multiplied by its velocity but this is not the definition
of momentum, as some people would have you think, and in fact
is only an approximation for nonrelativistic speeds. In nonrelativistic physics, momentum is a vector.
One of the most fundamental known laws of physics is that
momentum is conserved, meaning that the total amount of it
always stays the same. For example, if my dog runs into me and
I stop her motion, her momentum vanishes, but I recoil with the
same amount of momentum, so that the same amount still exists.
In the photo, the sprinters initially have zero momentum. Once
they kick off from the starting blocks, they have momentum, but
the planet earth recoils at a very small speed in the opposite
direction, so that the total is still zero.

As a measure of oomph, however, momentum cant be the

whole story. The firework in the photo has pieces that move in all
directions symmetrically. Since momentum is a vector, the momenta of the pieces cancel out. We want some other number that
measures how much of a bang was stored up in the gunpowder



and released as heat, light, sound, and motion when the gunpowder exploded. This is called energy, and its the E in E = mc2 .
In nonrelativistic physics, energy is conserved, and its conservation is essentially the only way of defining it operationally. For
example, if we define a unit of energy as a certain amount of heat,
then the energy content of a battery can be determined by using
it to heat something. Whatever amount of energy was gained as
heat, that same amount must have been lost by the battery.
In nonrelativistic physics, energy is a scalar. For example, if
we rotate a car, there is no change in the amount of energy stored
in its gas tank.
Kinetic energy is the term for the energy that a material object has because of its motion. Although I intend this to be a
book with almost no equations, it will be helpful to introduce an
equation for kinetic energy. Since conservation laws are additive,
kinetic energy must be proportional to the mass of an object.
For example, if two cars are moving at the same speed, but one
has twice the mass, it must have twice the energy. Experiments
at nonrelativistic speeds show that kinetic energy is also approximately proportional to the square of the velocity, i.e., the velocity
multiplied by itself. So for example if we drive a car twice as fast,
it has four times the kinetic energy. (This makes driving at high
speeds more dangerous than people intuitively expect.) Finally,
if we measure mass, velocity, and energy in their metric units
of kilograms, meters per second, and joules, then there is also a
conventional proportionality factor of 1/2. Putting all the factors
together, we have the formula K = (1/2)mv 2 . The presence of
the square is also connected to the fact that energy is a scalar. If
we pick a certain direction and call that positive, then velocities
in the opposite direction are considered negative. But driving a
car in the opposite direction doesnt negate its energy if it did,
you could fill your gas tank back up by driving that way! Mathematically, when we multiply a negative number by itself, we get
a positive result, so the kinetic energy comes out the same.

5.3. LIGHT




As a physics student, Einstein was taught that the universe consisted of two ingredients: matter and light. Because we cant
touch and feel light, and cant manipulate it with our hands,
its properties can be hard to pin down. But it often suffices to
work with some model of light. The ray model is the simplest.
Rays of light were probably one of the things Euclid had in mind
when he imagined perfectly straight lines. The wave model, discussed in section 12.3, p. 158, has the advantage of providing a
straightforward explanation of color: the different rainbow colors
are different wavelengths of light. The particle model seems to
contradict the wave model, but in fact can be reconciled with it;
today light is believed to have both wave and particle properties.

In the ray model of light, the only way your eyes can see
anything is if a ray goes into your eye. When you look directly
at a luminous object such as a candle flame, rays travel straight
from the flame to your eye. When you look at a nonluminous
object such as your hand, rays from a source of light such as the
sun hit your hand, are reflected, and then enter your eye.
Although physicists had good descriptions of both light and
matter before Einstein, there were subtle contradictions between
them, which bothered him as a young student. In the following
chapter this will lead us directly to the famous E = mc2 .



Chapter 6

E = mc2
We now trace essentially the same chain of reasoning that Einstein used in arriving at his most famous equation.


Energy shift of a beam of light

A ray of light always moves at c, and this leads to some big

differences in behavior between the energy of matter and the
energy of light. If a boxer throws a harder punch, his fist has
more energy because of its greater speed. But we cant make a
flashlight brighter by making its beam come out faster.
This leads to the following relativistic question, which turned
out to be crucial to Einstein. Suppose we move toward an oncoming beam of light at a significant fraction of c, as in a head-on
collision between two cars. What happens to the beams energy
when we switch to this frame of reference?
Since its speed is the same in the new frame, we might imagine
that its energy was unchanged as well. After all, if you dont
change the speed of a baseball or a bullet, you dont change its
energy. But this cant be right, for the following reason. The
law of conservation of energy is delicately constructed so that if
energy is conserved in one frame of reference, its conserved in
other frames as well. In a typical energy-transformation process,


CHAPTER 6. E = M C 2

such as the emission of a beam of light by a flashlight, we have

a bunch of different forms of energy before the transformation,
and a similar list afterwards. The sum of all the initial forms of
energy has to equal the sum of all the final forms. If we switch
to a different frame of reference, the numerical values of many
of these numbers change. (For example, if the flashlight was at
rest in the original frame, with zero kinetic energy, then in the
other frame, where its moving at a substantial fraction of c, it
has more energy than a nuclear warhead.) If energy is to be
conserved in the new frame as well, then all of the various forms
of energy must be different in the new frame, and the differences
must be such that the books still balance. There is no way to
make this work if we insist that a beam of light keep the same
energy in all frames.
So a beam of light must be brighter to an observer rushing
into the beam. How much brighter? The effect turns out to be
that the energy gets multiplied by the stretch factor introduced
in section 3.4, p. 50. For example, if the observer moves at 3/5
of c, then the stretch factor equals 2, and the beam has twice as
much energy. This use of the stretch factor is pretty much the
only possibility that makes sense, because we could have a chain
of observers, with B in motion relative to A, and C in motion
relative to B. When we combine velocities like this, the stretch
factors multiply, but so do the intensity factors for a beam of
light. For example, if B says the beam is twice as bright as A,
and C says its twice as bright as B, then C says its four times
as bright as A.
By the way, we do see this effect in real life. Because of the
expansion of the universe, distant galaxies are moving away from
us at speeds comparable to c. This makes their light dimmer and
harder to detect through a telescope. (There is also a color shift
toward the red end of the spectrum, which in the wave model
of light is described as a lengthening of the wavelength. This
wavelength effect is called the Doppler shift.)




Einsteins argument

We now have all the necessary ingredients to go through Einsteins argument leading to E = mc2 , and to interpret what the
equation means. In Einsteins 1905 paper on this topic, he gave
an argument using algebra, but Ill present it using a numerical
example instead.
Suppose that in Alices frame of reference A, we have a batterypowered lantern that is initially at rest. The lantern is set up so
that it shoots one beam of light to the right, and simultaneously
another beam of equal intensity to the left. We turn the lantern
on and then back off, and each of the two escaping beams of light
carries away 0.5 units of energy. Although the beams of light do
have momentum, momentum is a vector, so the momenta in opposite directions cancel out, and the lantern doesnt recoil. Since
the lantern is still at rest after the emission of the beams, its
kinetic energy both before and after emission is zero. Therefore
by conservation of energy, the 0.5 + 0.5 = 1 unit of energy in
the light must have come from the dissipation of 1 unit of energy
that had been stored in the battery.

Now suppose that this same process is observed by Betty, who

is moving to the right at 3/5 of c relative to Alice and the lantern.
In Bettys frame B, the leftward beam is increased in energy by
a factor of 2, while the rightward beam has its energy cut in
half. Their energies in frame B are 0.25 and 1.0 units. But now
we have a problem making the books balance on conservation of
energy. The lantern has some kinetic energy, but we dont expect
it to change its mass or velocity when the light is emitted, so we

CHAPTER 6. E = M C 2


would think the kinetic energy would remain unchanged between

the initial and final state, and we would be able to ignore it for
purposes of conservation of energy. Initially, we have:
1 unit of energy stored in the battery
After emission, we have:
0 energy stored in the battery
+0.25 units of energy in the right-going beam
+1.00 units of energy in the left-going beam
We have a discrepancy of 0.25 units. If we dont patch things
up somehow, conservation of energy will be violated. That is
of course logically possible, and there were in fact a couple of
occasions in the early twentieth century in which physicists did
seriously consider abandoning energy conservation as a fundamental law of physics. Einstein decided to do something trickier.
He decided to redefine energy. He proposed the following:
Equivalence of mass and energy
Mass and energy are equivalent, and the conversion between mass and energy units is given by the equation E =
mc2 .
Lets see what happens when we apply this idea to the example of the lantern. Initially, the battery contains 1 unit of
energy. This is equivalent to an amount of mass m = E/c2 .
In metric units, where c is a huge number, this comes out tiny.
For example, 1 joule of energy would be equivalent to about
0.00000000000000001 kilograms (or 1017 kg in scientific notation). But when we switch to Bettys frame, the lantern is moving at a very high speed, so this miniscule amount of mass could
end up contributing quite a bit of kinetic energy. How much extra kinetic energy do we get? We cant immediately answer this
question because although we have a formula for kinetic energy
(p. 74), that formula is only a nonrelativistic approximation. The
speed of 3/5 of c that were dealing with in our numerical example
is pretty high, so we shouldnt expect the nonrelativistic formula



to work exactly. If we nevertheless go ahead, take this mass and

the velocity of (3/5)c, and plug them into the formula, we get
an energy of 0.18 units, which is close to, but not exactly equal
to, the 0.25 units that we need in order to make conservation of
energy work. An advantage of Einsteins use of algebra over our
numerical approach is that his method makes it easier to demonstrate that at low velocities, where the nonrelativistic formula for
kinetic energy is known to be a good approximation, the discrepancy in conservation of energy is eliminated. It is eliminated at
higher velocities as well, but only if we use the correct relativistic
equation for kinetic energy.


Correspondence principle

The correspondence principle plays a crucial role here. We get

to rewrite the rules, redefining energy, but our new rules have to
be approximately consistent with the old ones in the situations
where the old rules had already been checked by experiments.
The figure shows an example. The match is lit inside the bell
jar. It burns, and energy escapes from the jar in the form of
light. After it stops burning, all the same atoms are still in the
jar: none have entered or escaped. The burnt match is lighter,
but only because some of its atoms are now in the form of gases
such as carbon dioxide. The gases were trapped by the bell jar,
so they still contribute to the total weight. The figure shows
the outcome expected before relativity, which was that the mass


CHAPTER 6. E = M C 2

measured on the balance would remain exactly the same. Mass

is conserved.
Now Einstein claims that the energy of the light escaping from
the jar is equivalent to some mass. How can this be? Experiments
like this had been done by physicists as far back in history as
Lavoisier, at the time of the French Revolution. But recall our
numerical estimate above. A joule is actually a pretty reasonable
estimate of the energy released by a burning match, and we saw
that a joule is only equivalent to an extremely small amount of
mass. Lavoisier was right, to within the precision of the balances
he had available. He simply didnt have a scale that measured
weights to seventeen decimal places.


Mass to energy and energy to mass

In Star Treks transporter technology, people are converted to

beams of energy and then reconstituted in another place. Before
Einstein, mass and energy had been thought to be separately
conserved, so this would be impossible. According to relativity,
they are equivalent, and we only have conservation of the total
amount of mass plus energy (with the two quantities converted
into compatible units before addition). The sum is called massenergy. Its as though we believed that Euros and Swiss Francs
were completely different things, until one day we found out that
there was a money changing shop that would let us trade in one
for the other.
Although nobody has yet succeeded in teleporting an object,
we do observe processes in nature that convert mass to energy and
energy to mass. The most common examples involve the forms
of radioactivity that are present naturally in our environment.
Some naturally radioactive substances in the earth produce
positrons, which are like electrons but have the opposite electric
charge. A form of antimatter, positrons annihilate with electrons
to produce gamma rays, a form of short-wavelength light.
Positron annihilation forms the basis for the medical imag-



ing technique called a PET (positron emission tomography) scan,

in which a positron-emitting chemical is injected into the patient
and mapped by the emission of gamma rays from the parts of the
body where it accumulates. Because the gamma rays fly off back
to back, the detector can reconstruct the line along which they
flew out and therefore the direction to the source of the radioactivity. In a typical application of PET, a patient with cancer is
given a form of radioactive glucose, which emits positrons. The
glucose tends to become concentrated in the cancerous tissue,
allowing an image of the tumor to be reconstructed.
Conversion of energy to matter also occurs naturally. For
example, we are naturally surrounded not just by radiation from
minerals in the earth but also by radiation from outer space,
called cosmic rays. Some of these cosmic rays are gamma rays,
which can interact with matter by converting part of their energy
into the creation of an electron and a positron.


The energy-momentum vector

We saw earlier that spatial displacement wasnt a valid relativistic vector, and to make a valid one, we had to combine it with
a time interval. A similar approach works for mass-energy and
momentum. To get a valid relativistic vector, we have to combine


CHAPTER 6. E = M C 2

mass-energy with momentum.

The figure shows how this plays out in a couple of examples
involving different frames of reference. A point on the graph
represents a thing with a certain energy and a certain momentum.
The lower dot represents a beam of light, and logically enough,
it lies on the light cone. That is, its energy and momentum are
numerically equal, if we choose units in which c equals 1. In
terms of ordinary metric units, we have to include c as a conversion factor, and the momentum of a beam of light equals its
energy divided by c. Because c is a big number in metric units,
the result is a small momentum, and this is why a flashlight
doesnt kick like a gun.



In the frame of reference represented by the white grid, the

beam of light has an energy of 2 units and a momentum of 2
units. The black grid is the frame of reference of an observer
chasing the beam at 3/5 of c. To this observer, the beam has
energy and momentum both equal to 1.
The upper dot represents a rocket ship. The black grid represents the frame of reference of an observer aboard the ship. To
that observer, the ship is at rest. By symmetry, its momentum
can point neither to the right (positive) nor to the left (negative), so it must be zero, and zero is the value we read off of the
graph. In the ships own frame, it isnt moving, so it has zero
kinetic energy. But its mass-energy isnt zero, its 4 units. The
ships mass is 4 units, and in this frame thats the only kind of
mass-energy it has.
In the white frame, the ship has 3 units of momentum and 5
units of energy. This energy is due to both the ships 4 units of
mass and an additional 1 unit of kinetic energy.
Since the ships energy-momentum vector is timelike, by definition its magnitude equals its size as measured in the frame
where its at rest. Its magnitude is 4 units, which is its mass.
The magnitude is the same in all frames of reference, so an observer in the white frame of reference agrees that its mass is 4.
This is how we define mass in relativity: as the magnitude of the
energy-momentum vector.
The beam of light has an energy-momentum vector that is
lightlike, so its magnitude is zero. This is why we say that a
beam of light has zero mass. Again, all observers will agree on
this fact.
In general, matter is made out of particles whose energymomentum vectors are timelike, because they have nonzero mass.
Light, as well as certain more exotic creatures such as gluons
and gravitational waves, has zero mass and therefore a timelike
energy-momentum vector. A hypothetical third category of particles, called tachyons, would have a spacelike energy-momentum
vector. Tachyons probably dont exist, which is a shame because
the universe would be a wilder place if we had them. In section


CHAPTER 6. E = M C 2

6.6, p. 88, Ill say a little about searches for tachyons and what
tachyons might be like.
Money isnt much fun if you cant spend it, and energymomentum isnt much fun if you dont ever change it from one
form to another. The first figure below shows a collision between two pool balls, with a diagram showing how their energymomentum vectors add up. As actually happens in this kind of
head-on shot (unless you use a lot of spin on the cube-ball), the
cue ball stops dead, transferring all of its motion to the ball it
hits. (The diagram is actually a little unrealistic, because Ive
drawn the balls moving at about 20% of c, which would not only
destroy them on impact but also kill everyone within a considerable radius.)

One way to check that energy-momentum is conserved here

is to set up each pair of vectors tip-to-tail as in the method of
vector addition shown in section 4.2, p. 64. An easier way to
verify it is that the two sides of the equation are the same except
that the order of the things being added up has been reversed;
vector addition doesnt care about the order, just as in ordinary
addition, where 1 + 3 is the same as 3 + 1.
Most of what we see in this diagram is simply the mass of the
balls. The moving ball does however have a little bit of energy
(extra height) and some momentum (leaning to the right, so that



its spacelike part isnt zero).

Now heres a diagram of a somewhat more realistic example,
one that really requires relativity. This is a head-on collision
of two subatomic particles called quarks, such as occurs at the
Large Hadron Collider near Geneva, Switzerland. The quarks
are actually parts of larger particles protons but on this
scale we only see one quark from one proton colliding with one
quark from another proton. We dont actually know if quarks are
pointlike or have a finite size, but Ive drawn them here as having
some bulk to them, and as ovals rather than circles to represent
the effect of Lorentz contraction. In this example, some of the
energy of the incoming quarks happens to produce something
called a Higgs particle. That is, energy has been converted into
matter via E = mc2 . The Higgs particle was predicted to exist in
the 1960s, and the LHC finally confirmed its detection in 2012.
When the LHC was being designed, it was expected to be almost
certain to find the Higgs. In fact, the fear was that the Higgs
was the only thing it would find at all, and that appears to be
what has happened. If so, then many particle physicists will be
very depressed, but no doubt they will use it as a justification for
another project, even larger and more expensive than the LHC.



CHAPTER 6. E = M C 2

Why cant I have tachyons?

The energy-momentum vector is timelike for matter and lightlike

for massless phenomena such as light. A hypothetical third category of particles, called tachyons (TACK-y-ons), would have a
spacelike energy-momentum vector. They would always go faster
than c, and their mass would be an imaginary number, like the
square root of 1.
In section 3.5, p. 51, we saw both theoretical and experimental evidence that no continuous process of acceleration can boost
a material object past c. That didnt, however, address the question of whether one could surpass c through some discontinuous
process, such as the jump to hyperspace in Star Wars. This
loophole now appears to be closed off. We observe that mass is a
permanent, fixed property of a material object, and it therefore
seems that a material object like the Millenium Falcon could not
go faster than c, or else it would have to be transformed into a
cloud of tachyons.

But that doesnt mean that tachyons dont exist. If they did,
it would be cool. Consider the figure above. In the frame of
reference represented by the white grid, someone uses a burst of
tachyons to send a signal from event A to event B. According



to this observer, B is in As future, but so far away that there

would have been no way for anything to get from A to B except
by going faster than c.
But according to the observer represented by the black grid, B
is in As past, since B happens at a time less than zero (i.e., before
the time arbitrarily chosen as a reference), while A happens at a
time greater than zero.
This means that if tachyons exist, we can automatically use
them to send signals back in time. (Personally, Id tell myself to
sell all the stocks in my retirement account in 2007, then buy in
But do they exist? This is a question that can only be answered by searching experimentally. The most obvious experimental signature of tachyons would be motion at speeds greater
than c. Negative results were reported by Murthy and later in
1988 by Clay, who studied showers of particles created in the
earths atmosphere by cosmic rays, looking for precursor particles that arrived before the first gamma rays.
One could also look for particles with spacelike energy-momentum vectors. Alvager and Erman, in a 1965 experiment, studied the radioactive decay of thulium-170, and found that no such
particles were emitted at the level of 1 per 10,000 decays.
Some known subatomic particles, such as neutrinos, dont
interact strongly with matter, and are therefore difficult to detect directly. Its possible that tachyons exist but dont interact
strongly with matter. If so, then it might be possible to infer
their existence indirectly through missing energy-momentum in
nuclear reactions. This is how the neutrino was first discovered.
An accelerator experiment by Baltay in 1970 searched for reactions in which the missing energy-momentum was spacelike, and
found no such events. They put an upper limit of 1 in 1,000 on
the probability of such reactions under their experimental conditions.
For a long time after the discovery of the neutrino, very little
was known about its mass, so it was consistent with the experimental evidence to imagine that one or more species of neutrinos


CHAPTER 6. E = M C 2

were tachyons, and Chodos et al. made such speculations in

1985. A brief flurry of reawakened interest in tachyons was occasioned by a 2011 debacle in which the particle-physics experiment OPERA mistakenly reported faster-than-light propagation
of neutrinos; the anomaly was later found to be the result of a
loose connection on a fiber-optic cable plus a miscalibrated oscillator. An experiment called KATRIN, currently nearing the
start of operation at Karlsruhe, will provide the first direct measurement of the mass of the neutrino, by measuring very precisely
the missing energy-momentum in the decay of hydrogen-3.

Part III



Chapter 7

Newtonian gravity

Galileo and free fall

Galileo had an astounding knack for zeroing in on scientific issues

that we still consider crucial four centuries later. Special relativity can be traced directly back to his insistence on the principle
of inertia, as opposed to the Aristotelian idea that motion naturally stops if no force is applied. A similar Galileo-Aristotle
controversy holds within it the seed of general relativity.
Aristotle thought that an object falling in air was analogous
to one sinking in water, and claimed that objects fell at a certain
speed, which would be proportional to the objects weight and
would also depend on both the object and the medium through
which it was falling. In some ways this is a reasonable description,
but its very complicated and in some ways inaccurate. Its both
simpler and more fundamentally important to consider the case
in which the frictional force from the surrounding medium is negligible, which is what happens when the density of the medium
is low, and the falling object is relatively dense and heavy. You
might want to give this a try yourself right now. You probably
have some handy objects of contrasting weight, such as a coin
and a shoe. Stand up and release them side by side.
My own shoe is about 50 times heavier than the nickel I had



handy, but it looks to me like they hit the ground at exactly the
same moment. So much for Aristotle! Galileo, who had a flair
for the theatrical, did the experiment by dropping a bullet and a
heavy cannonball from a tall tower1 (panel 1 of the figure).

Nor does a given object fall with a single definite speed. Starting from rest, it speeds up and speeds up. The ingenious Galileo,
who had only primitive clocks, convinced himself of this by doing
experiments with balls rolling down inclined planes. By extrapolation, we expect the same to hold when the plane is tipped
all the way up to the vertical, so that the object is falling freely
(panel 2). In modern units, it turns out that objects on earth
gain very nearly 10 meters per second (m/s) of speed in every
second, so that if we release a rock at rest (0 m/s), then after
one second of falling its speed is 10 m/s, after two seconds 20
m/s, and so on. This rate of speeding up is called the rocks
acceleration, and it is 10 meters per second every second, i.e., 10

His contemporary and first biographer Viviani says that it was the Leaning Tower of Pisa, although some historians are skeptical about this.



meters per second per second, also written as 10 m/s2 .

Not only did Galileo disprove Aristotle empirically, but he
also pointed out a logical contradiction in Aristotles own reasoning. Aristotle said that heavier objects fell faster than lighter
ones (panel 3). If two rocks are tied together, that makes an
extra-heavy rock, which should fall faster, 4. But Aristotles theory would also predict that the light rock would hold back the
heavy rock, resulting in a slower fall, 5. In reality, we find that
all of these objects fall with the same acceleration, 6.
This universality of free fall turns out to have surprisingly farreaching consequences. Eventually, starting from little more than
this fact, we will see that gravity can be described as not being
a force at all, but rather a warping or curvature of spacetime.
Free fall includes more possibilities than falling straight down.
If we watch Galileos experiment in a frame of reference moving
vertically or horizontally, we may see the balls as going up rather
than down, or flying along curved arcs. Therefore projectiles
must also have universal motion, in the absence of friction. The
figure below shows computer simulations of the trajectories of
a golf ball, baseball, and softball, all shot at the same initial
speed and angle. (The baseball hit in this simulation goes a
distance of 200 meters, roughly a home-run distance.) Although
the trajectories of the three types of balls are not the same in air,
their trajectories would be exactly the same (and considerably
longer) in vacuum.

The universality of free fall has been tested in many different

experiments over the years. In a silly educational demonstration,
Apollo 15 astronaut David Scott dropped a feather and a geologists rock hammer side by side. Online, you can watch video



footage in which the two objects fall to the lunar surface side by
side. One can vary not just the weight but also the composition
of the falling body, verifying, for example, that a brass ball and
a lead ball fall with the same motion. It was realized early on
by Galileo himself, in fact that the same fundamental issue could be probed by experiments that appeared superficially
very different and that allowed higher precision. For example, if
gravity produces different accelerations in free-falling balls made
of brass and lead, then pendulums with bobs made of brass and
lead should also oscillate at different rates. Experiments of this
type were done by E
otvos around 1900, and later refined by Braginskii, Dicke, and their co-workers to give a precision of about
one part in 100 billion. Attempts are even under way to test
whether the universality extends to antimatter. Its conceivable
that antimatter is actually repelled by the earth, so that it would
fall up, like the imaginary upsydaisium from the Rocky and
Bullwinkle cartoon series.



You can count yourself lucky to live in a universe where free fall
seems to be universal. At this moment, both you and the planet
earth are in a state of free fall, being pulled in by the suns gravitational attraction. As a result, Aristotle would have expected
the earth to accelerate toward the sun much more violently than
your body, with the result that the moment you stepped outside,
the world would fall out from under you and you would go flying
off into outer space. Ouch!
The earth revolves around the sun in an approximately circular orbit. On p. 19 we saw that circular motion requires a force;
in the figure on the facing page, if the string breaks the rock will
fly off in a straight line. Similarly, the suns gravitational force
is what keeps the earth from zooming away.
Understanding the motion of the planets was originally considered important partly because in ancient Europe, one of the





best ways to make a living doing astronomy was to cast horoscopes at the time, there was no clear distinction between the
science of astronomy and the pseudoscience of astrology. After
the Copernican-Galilean revolution (see section 2.7, p. 44), the
planets were described as revolving around the sun, and the observational astronomer Johannes Kepler (1571-1630) arrived at
a complete and fairly accurate mathematical characterization of
their motion in terms of three laws. Two of these laws describe
the speeds of the planets motion, while the remaining one states
that each planets orbit is an ellipse, with the sun lying at a
certain well-defined off-center point called a focus. One way of
defining an ellipse is that its a circle that has been distorted
by shrinking and stretching along perpendicular axes. An ellipse
can also be constructed by tying a string to two pins and drawing a curve as shown in the figure, with the pencil stretching the
string taut. Each pin constitutes one focus of the ellipse. Just as
a square is a special type of rectangle, a circle is a special case of
the ellipse. In a circle, the two foci coincide.


Newtons laws

Newton firmed up Galileos work on motion by defining a set of

natural laws that applied both to objects on earth and to the
heavens a radical idea at the time. Well soon see that these
laws give a simple and natural explanation for the universality of
free fall. Newton was also able to use his laws to prove all three
of Keplers laws for the motion of the planets, although this proof



is a longer and more mathematical, and we wont go into it here.

In Newtons description, objects push and pull on each other
through attractive and repulsive forces. Gravity is a specific type
of force, a universal attraction between any two objects that is
proportional to their masses. When objects interact through a
force, each ones momentum is changed in fact, we can define
force as the rate of change of momentum. The changes in momentum are like transferring money from one bank account to
another; the total amount of momentum is conserved.

When an object has more inertia more mass it takes

less motion to express the same amount of momentum. For example, if I throw a golf ball at an empty can, the can recoils
violently, while the same hit on an aircraft carrier produces an
imperceptibly small recoil. The momentum absorbed by the can
and the aircraft carrier are the same. In the figure, a professional
football player collides head-on with a little old lady. Momentum
is conserved, so whatever momentum she gains is just enough to
balance out what he loses. Therefore their forces on each other
are equal. But the effects on their motion are not equal. Her
mass is much less, so the change in her motion is much more



In summary, Newtons description of motion consists of the

following three laws:
1. If no force acts on an object, its momentum doesnt change.
2. If forces act on an object, its momentum changes at a rate
given by the total force.
3. Forces always occur in pairs. If object A exerts a force on
object B, then B is also exerting a force on A. These forces
are in opposite directions and have equal magnitudes.
Newtons first law is a special case of the second law. The third
law is essentially a statement of conservation of momentum; the
rate at which A loses momentum to B equals the rate at which
B gains it from A.
Newtons laws give an explanation for the universality of free
fall. If my shoe is 50 times more massive than a nickel, then the
earths gravitational force on the shoe is 50 times greater. But
the shoes inertia is also 50 times greater, so that when I drop
shoe and coin together, their responses to the force are exactly
the same. But Newtons explanation is a sort of non-explanation
explanation. It depends on the fact that an objects inertia is
proportional to its gravitational mass. That is, we have two
separate concepts of mass, and Newton cant tell us why theyre
equal. He just observes that they are. When I shake a lead brick
back and forth, I sense its inertia, or inertial mass. When I put it
on a scale, I measure its gravitational mass. There is no known
reason why these two things have to be related; it just seems that
they are. Besides gravity, we have other fundamental forces, such
as electricity. These forces have no such property.


Tidal forces

If the shoe and the coin start out right next to each other, they
stay right next to each other as they fall. A different situation
is shown in figure 1. In this spacetime diagram, objects A and



B start out at unequal distances from the earth. Because A

is closer, it feels a stronger gravitational force, so it accelerates
more rapidly. The result is that A and B become more and more
separated from each other.

Figures 2 and 3 show a spectacular example of this effect. In

1994, a comet named Shoemaker-Levy 9 approached the planet
Jupiter on a collision course. A comet is a loosely packed dirty
snowball, so there is not much holding it together. As this comet
approached Jupiter, the small difference in Jupiters gravity between the front and back of the comet was enough to pull it apart,
and the fragments subsequently separated from each other even
more, forming the string of pearls shown in figure 2. The fragments hit Jupiter in succession, and because Jupiter spins on its
own axis, the impacts occurred at different places in the jovian
cloud-tops. Figure 3 shows the resulting string of bruises, each
about as big as the planet earth.



Chapter 8

The equivalence

The equivalence principle

The universality of free fall has the following very subversive

consequences. Consider Alice, the girl in figure 1. She is shown
in a cutaway view, sealed inside a well-lit and air-conditioned
box, which happens to be her favorite place to go and think
about physics. (Its small, but its private and its all hers.) She
feels pressure from the seat of the chair, and this would normally
suggest that she was experiencing a gravitational force, perhaps
from a nearby planet (figure 2). But Alice cant infer this without
peeking outside at her surroundings. Its equally possible that the
box is in deep space (figure 3), with no gravity whatsoever, but
is being towed by a rocket ship, so that it accelerates constantly.
No possible experiment done inside the box can tell her which of
these is the case.
For example, suppose that she drops a shoe and a coin side
by side. If shes really experiencing gravity, then due to the
universality of free fall, they will hit the floor at the same time.
But this is exactly the same observation she would make in the
rocket-tow scenario: the shoe and coin would continue moving



forward at constant speed, but the accelerating floor would catch

up with them and hit them both at the same time.
From the perspective of Newton and Galileo, this is deeply
disturbing. According to one scenario, the box is an inertial
frame of reference. Based on the other, its not.
These ideas lead relativists to the following point of view.
Gravity is not really a force. Any observer who thinks shes
experiencing a gravitational field can equally well interpret her
observations by switching to a different frame of reference, a freefalling frame of reference, in which there is no gravity. These
free-falling frames are the ones that deserve to be called inertial.
(By this definition, the room in which youre reading this book is
not an inertial frame.) Summarizing these observations, we say
that the existence of a gravitational field is equivalent to choosing
an accelerated frame of reference. This is called the equivalence
principle, and it is the single most important feature of Einsteins
theory of general relativity.
The equivalence principle has an important application to
flying an airplane. The pilot of a plane cannot always easily tell
which way is up. The horizon may not be level simply because
the ground has an actual slope, and in any case the horizon may
not be visible if the weather is foggy. One might imagine that
the problem could be solved simply by hanging a pendulum and
observing which way it pointed, but by the equivalence principle
the pendulum cannot tell the difference between a gravitational



field and an acceleration of the aircraft relative to the ground

nor can any other accelerometer, such as the pilots inner ear. For
example, when the plane is turning to the right, accelerometers
will be tricked into believing that down is down and to the
left. To get around this problem, airplanes use a device called an
artificial horizon, which is essentially a gyroscope. The gyroscope
has to be initialized when the plane is known to be oriented in a
horizontal plane.1


Solving the chicken-and-egg problem

of inertial frames

An advantage of equivalence principle is that it allows us to

say for sure whether a certain frame of reference is inertial. In
the Newton-Galileo theory, there was a crippling chicken-andegg problem with this. For example, on p. 17 we visualized the
frames of reference of a cow and a passing car. Suppose were
already convinced that the cows frame is inertial. The cow sees
the car moving along in a straight line, at a speed that appears
to be constant in the cows frame. Therefore the cow says that
the cars motion is inertial the car isnt going over bumps,
No gyroscope is perfect, so over time it will drift. For this reason the
instrument also contains an accelerometer, and the gyroscope is always forced
into agreement with the accelerometers average output over the preceding
several minutes. If the plane is flown in circles for several minutes, the
artificial horizon will be fooled into indicating that the wrong direction is



running into a tree, or anything like that. Similarly, if the driver

is already convinced that the cars motion is inertial, then she
can tell that the cow, who to her is moving backward at constant
speed, also has a good inertial frame.
But now perhaps youve noticed the difficulty. Given one
inertial frame of reference, we can tell which other frames are
also inertial. But how do we get started? Similarly, imagine that
you were in Tanzania, and needed to make sure that the money
you were handling was valid currency. If you had a 1000 shilling
note from a reputable source, you could compare other bills with
it and make sure they looked the same. But without a knowngood example to start with, youd have no way of telling whether
shopkeepers were all handing you Monopoly money as change.



Lets say the girl Alice in the box on p. 104 wants to know
whether shes in an inertial frame. To do this, she needs to know
whether forces are causing her to accelerate. She knows about
the equivalence principle, so she doesnt consider gravity to be
a force. Therefore the question of whether gravity is acting on
her doesnt even arise. She does notice, however, that the seat
of the chair is touching her and making a force on her. She
knows that this force is accelerating her, so she knows her frame
is noninertial.
For an example of a frame that we know is inertial, imagine
building a box with layers of shielding, as suggested in the figure on the preceding page, where the box is free-falling in outer
space. One layer of shielding is aluminum foil to keep out any
external electric forces. Another layer provides shielding against
magnetism. (There are special products sold for this purpose,
with trade names such as mu-metal.) We shield against all the
forces except for the force of gravity, we dont consider to be a
force. Now inside the box we introduce a black ball and a white
ball. The roles played by the two balls are completely symmetrical, but if we like, we can think of one ball as playing the role of
an observer. For this reason Ive drawn a ladybug clinging to the
white ball. The bug can, if she likes, define herself to be at rest,
so that in her frame of reference only the black ball is moving.
She knows shes shielded from all forces, so she knows her frame
of reference is inertial. The chicken-and-egg problem is solved.
Once she knows that her frame, tied to the white ball, is inertial,
she can determine that the black balls motion is inertial as well.


Apparent weightlessness

This may all seem pretty bizarre, so lets connect it to the real
world. At an amusement park near where I live in Southern
California, there is a ride called the Supreme Scream. Riders are
raised in chairs up to the top of a tower, then dropped toward the
ground in free fall. When I tried this ride, I got a queasy feeling



as though my guts were floating rather than staying tied down

probably the same sensation that the Apollo astronauts got
when they experienced weightlessness by traveling very far from
the earth. A popular practice among people who get on this ride
is to take a coin and release it in the air in front of them after the
free fall has begun. An observer on the ground describes the coin
as falling along with the rider, but to the rider, the coin appears
to float weightlessly. The rider is doing a local experiment in an
inertial (free-falling) frame of reference. In this frame, gravity
doesnt exist.
The Supreme Scream is 95 meters tall, which limits the period
of free fall to about four seconds. If we want to experience apparent weightlessness for longer we have to fall from higher. For
example, NASA maintains a plane, affectionately known as the
Vomit Comet, for training purposes. The plane climbs steeply
and then free-falls along an trajectory that in the earths frame
is an arc of a parabola. The rise and fall together last about 25
seconds. The photo shows physicist Stephen Hawking on a ride
in the Vomit Comet.



Exactly the same thing is going on for astronauts aboard the

International Space Station. As in the case of the Vomit Comet,
their free fall carries them not straight down but along an arc
(according to an observer in the earths frame). Because theyre
moving at such a high speed (about eight kilometers per second),
the arc is a very broad one broad enough to carry them around
the world in a circular orbit, without ever hitting the ground.



The equivalence principle guarantees that we can always define

an inertial frame of reference, by choosing a free-falling object as
a point of reference. It would be convenient if we could define a
single inertial frame big enough to cover the entire universe, but
we cant. In the figure, two girls simultaneously drop down from
tree branches one in Los Angeles and one in Mumbai. The
girl in LA defines an inertial frame, but if she tries to expand
her frame of reference to cover her counterpart in Mumbai, the
Mumbai girl does not appear to have inertial motion: to the
LA girl she looks like she is falling up (with twice the normal
acceleration of gravity as viewed in an earth-fixed frame).

This example shows that the equivalence principle is a purely

local thing. Locally, we can always adopt a free-falling frame of
reference in which gravity doesnt exist, so that in this sense gravity is only an illusion. But we cant do the same thing globally,



i.e., across all of time and space. Gravity is after all inescapably
real, if we look at large scales. Similarly, tidal forces cannot be
dismissed as illusions; for example, they destroyed of the comet
in figure 2 on p. 101.


Gravitational time dilation

The equivalence principle makes powerful and surprising predictions. In figure 1, Amy (gray) stands on the ground while her
twin brother Bob (white) jumps straight up in the air next to
her. Because were using our standard graphical conventions for
spacetime diagrams, it looks like Bob is doing a long jump, but
remember, this is just a straight up-and-down jump the vertical axis here is time. If we didnt know about the equivalence
principle, wed expect that slightly less time would pass for Bob
than for Amy, since hes moving and shes not; wed think there
would be time dilation.
But the equivalence principle makes exactly the opposite prediction. Based on the equivalence principle, weve been led to
define free-falling frames as inertial. Therefore the more natural
depiction of the situation is the one in figure 2, where we adopt
Bobs frame of reference. Bob is not experiencing any forces (remember, gravity isnt a force), so his frame is inertial. According
to Bob, Amys frame is noninertial, because she does experience
a force from the floor on her feet. That force is accelerating her.
Her world-line is the one that is curved, not his. Based on the
principle of greatest time, he will experience more time than she
does. Less time will pass for her, not him.
A more practical version of this experiment was carried out
in 1976 by the Gravity Probe A experiment, in which an atomic
clock was launched straight up (relative to the ground) aboard a
space probe, to an altitude of about 10,000 km. As predicted by
the equivalence principle, the clock aboard the free-falling rocket
experienced a little more time (about half a part per billion) than
similar clocks on the ground.






Time dilation with signals

Gravity Probe A crashed into the Atlantic Ocean at the end

of its flight and was destroyed, so it couldnt be reunited with
the ground-based clocks and compared with them again side by
side. Instead, radio signals were sent back and forth between the
probe and the ground station. To understand how this works,
lets simplify the situation and go back to the scenario of Alice
in her noninertial box (p. 103).

Suppose that a light source on the floor of Alices box emits

two upward flashes of light, separated by a small interval of time,
so that one chases the other as they rise toward the ceiling, where
they can be detected. In an inertial frame, we say that the box
is accelerating upward. Therefore after the first flash arrives at
the detector, the continued acceleration of the box causes the
second flash to take a little extra time to arrive. It matters that
the box is accelerating, not just moving. Figure 1 shows the
case where the boxs motion is inertial rather than accelerated.
Flashes are emitted from the floor at A and B, and detected at
the ceiling at C and D. (You have to visualize the box tipped over
sideways.) The time intervals AB and CD are the same. But the
accelerated case, shown in figure 2, is different: GH is longer than
EF. Because the speed of light is big, this effect would be small,
but the time interval between the arrival of the flashes would be
slightly lengthened compared to the time between their emission.
If were willing to take the equivalence principle seriously,
then we should expect exactly the same effect in a laboratory
that sits in a gravitational field. That is what was observed in



a famous experiment by Pound and Rebka at Harvard in 1959.

Figure 3 above shows the basic idea. The experiment was done in
a 22 meter tower in one of the universitys lab buildings. Rather
than sending two discrete flashes of light up from the floor to
the ceiling, they used the fact that light is a wave, so that two
successive wave crests play the role of the two flashes. The light
waves used were gamma rays emitted in the decay of a radioactive
source. The figure below shows a simplified diagram of the setup,
and photos of the two experimenters at the top and bottom of
the tower.

The effect Pound and Rebka were trying to measure was ex-



tremely small. The time of flight for a gamma ray, moving at the
speed of light, to travel 22 meters is only about 70 nanoseconds
(billionths of a second). If we describe the tower in a free-falling,
inertial frame of reference, then its upward acceleration at 10
meters per second squared makes it move upward by only about
three millionths of a billionth of a meter (31014 m) during this
time of flight. The lengthening of the time interval between the
arrivals of the two flashes is thus extremely small, and detecting
it required several tricky techniques. The result was exactly as
predicted by the equivalence principle.
An observer in a free-falling, inertial frame of reference explains the results of the experiment as in figures 2 and 3 on
p. 112: we have two successive signals, and they travel at the
same speed c, but they have to travel unequal distances so the
reception of the second one is delayed.
But an observer in the noninertial frame of the earths surface
does not have this explanation available. In this frame, the tower
was not moving, so the signals traveled equal distances. This
observer is forced to conclude instead that time itself runs at
different rates at the top and bottom of the tower. This is similar
to the gravitational time dilation effect discussed in section 8.5.
The cobalt-57 nucleus in the Pound-Rebka experiment is acting
like a tiny clock. When it emits a gamma ray, the successive wave
crests are like the clock making two ticks. An observer at the top
of the tower, receiving the ticks, observes them to be abnormally
far apart. This is exactly what we would expect if the clock was
running slow because of a slower flow of time at the bottom of
the tower.
This may seem like a rash generalization, and indeed there
were probably quite a few physicists back in the 1950s who would
have expressed some doubt that a result like Pound and Rebkas
would also be observed with real clocks that count and measure
time units, as opposed to microscopic clocks that are merely
subatomic particles. But remember that such an effect was observed, for example, in 1978 by Iijima (p. 32) in an experiment
using one atomic clock at the base of a mountain and another



clock at the top.

Summarizing all of these examples, we conclude that time
runs more slowly when youre low down in a gravitational field,
faster if youre higher up. This summary also works as an alternative explanation for the thought experiment in which Bob
jumps while his twin sister Amy stands on the ground. Analyzing
figure 2 on p. 111, we previously said:
Bobs motion is inertial while Amys is not. Therefore by
the principle of greatest time, Amy experiences less time
than Bob.
An equally valid explanation, leading to the same numerical predictions, is:
Amy experiences less time for a combination of two reasons.
(1) As seen in Bobs inertial frame, she is the one who is
moving. Motion causes time dilation. (2) Amy is also lower
in a gravitational field than Bob is. For this reason as well,
Amy experiences less time. These two effects reinforce each
other, and their sum is the observed amount by which time
was slowed.
By the way, there seems to be a very common misconception
about gravitational time dilation, which is that it involves time
slowing down where gravity is stronger. Not true! Its height
that matters. For example, consider the scenario in figure 3 on
p. 104, where Alices box is being towed by the spaceship. If she
measures the fake gravity near the floor of the box and near
the ceiling, she will get exactly equal results. If she duplicates the
Pound-Rebka experiment inside her box, she will see an effect,
because of the difference in height. Or imagine sending a clock
to the center of the earth. At that location, the planets gravity
pulls equally in all directions, so by symmetry there is zero gravitational force. We would nevertheless find that this clock was
slow compared to one on the earths surface.



Chapter 9

Curvature of spacetime
This chapter is called curvature of spacetime, which sounds
pretty scary, but if youre willing to wholeheartedly accept the
equivalence principle, it isnt really that bad.


What is straight?

What do we mean when we say something is straight rather than

curved? Say Maria wants to check that a line on a piece of paper
is straight. Shell probably use a ruler. But how does she know
the ruler is straight? To check that the ruler was straight, she
might hold it up to her eye and sight along it. That amounts to
checking its straightness against the straightness of a ray of light.
Now how does she know that a ray of light is straight?
Well never get anywhere unless we go ahead and define something as straight to start with. Maria ended up resorting to a
ray of light. A ray of light is a thing that, in this example, we
assume was not being acted on by any force. In other words, the
best we can probably do is to define a straight line as the path
of an object that is moving inertially. That object could be a
flash of light from a laser, or it could be a baseball. And since
were talking about relativity, we want to define straightness in
the context of spacetime, i.e., we want to take this as a definition



of what it means for an objects world-line to be straight.

So straight just means inertial. The figure on the facing
page shows a realistic example. The woman fires her toy cannon,
which shoots a fist-sized steel ball bearing upward at about the
speed and angle of a home-run baseball.1 The spacetime diagram shows one dimension of time and one dimension of space
(the vertical one); the horizontal dimension has been suppressed,
which is why the ball looks like its going to hit her on the head
at the end. The diagram is constructed in an inertial frame of
reference that coincides with the initial rest frame of the ball. In
other words, the point of reference is a hypothetical projectile
shot just like the ball bearing, but for which air friction is negligible, so that no forces act on it. (Remember, we dont count
gravity as a force.) In this frame, the woman, the cannon, the
air, and the dirt all zoom downward, turn around, and then come
back up; this is sort of like the view youd see of the ground if
you were a bug riding on the perfectly inertial ball. Because of
air resistance, the balls trajectory is slightly curved; it is swept
downward by the air rushing down past it, and then decelerated
again as the air comes back up.
The principle of greatest time (section 2.2, p. 36) says that
of all the possible world-lines that a clock can follow from one
event to another, the one that gives the greatest elapsed time is
the inertial one. This leads to an analogy between straightness
in space and straightness in spacetime. In space, a straight line
gives the smallest distance between two points. In spacetime,
a straight world-line gives the largest proper time between two
events. The only change is that smallest switches to largest. The
reason for the switch is simple. Of the three world-lines in the
figure, the cannons takes the most ink to draw, and this means
that its world-line is the one that least resembles an inertial one.
A clock moving along with the cannon would therefore suffer the
most time dilation, and show the least elapsed time.

This was simulated for a steel ball of 4 cm diameter, shot at an angle of

40 degrees above horizontal and an initial speed of 45 meters per second.






The parable of the bugs

The close analogy between spatial straightness and straightness

in spacetime leads to the following parable, illustrated in the
figure. Two blind bugs get in an argument.
A Flock of Seagulls was the greatest band ever.
Oh, they were totally derivative. Id say Sonic Youth
The argument escalates. Words are said that cant be taken
back. Embittered, the bugs set off from A in different directions
to find new urban spaces with better brewpubs and more access
to sustainably grown kale. Unexpectedly, they run into each
other again at a drum circle in a public park in B.
What are you doing here?
Me? I went straight and never looked back.
Thats not possible. I went straight too, and we left A traveling in different directions.
Because Ive drawn the figure in a certain way, the explanation of the mystery seems obvious to us. But the bugs are blind
and cling to the surface of their world, so their mental picture
of their universe is two-dimensional. They can nevertheless work



out what has happened. Their two straight paths have formed a
kind of shape that you never learned about in kindergarten. The
shape with four sides is a square, three sides make a triangle, and
two is
Oh yeah, says one bug knowingly, its a lune. Really obscure shape, you probably never heard of it.
You cant draw a lune on a flat surface using straight lines.
If a lune exists, it means that the surface is curved. Whats
crucial is that the bugs can detect, understand, and measure this
curvature without knowing that their two-dimensional world is
embedded in a three-dimensional space. I could rewrite the story
so that the third dimension didnt actually exist in the bugs
universe, and all the events would be the same.
The bugs can even define numerically how curved their world
is. One good measure of curvature would be to take the angle
at the top or bottom of a lune, and divide by its area. This is
essentially2 the same as a measure of curvature called the Gaussian curvature. We can imagine making the bugs world less and
less curved by making the sphere bigger and bigger. As it gets
very big, its surface becomes very flat, the distance between A
and B gets very big, the area of the lune also gets very big, and
the Gaussian curvature gets very small (because dividing by a
big number produces a small result).
Although I presented this using a fairy-tale scenario, we can
encounter all of these concepts on the earths surface. Because
the earth is big, its curvature is small, and we dont notice in
everyday life that its not flat. A motorboat cruising without
turning, or an airplane flying straight and level, will trace a path
on the earths curved surface that is in some sense straight its
as straight as any path could be. Such a path is called a geodesic,
and the same term as also been coopted by relativists for use in
spacetime, where it describes the world-line of an object moving

Essentially means we dont worry about constant factors such as the

one that arises from choosing the units of measurement for the angle.




Gravity as curvature

The figure on this page shows the parable of the bugs translated from space into spacetime. Spacesuited astronauts 1 and
2 are initially flying off in opposite directions. Each astronaut
can verify that she is not being acted on by any forces gravity isnt a force and therefore knows that her world-line is
straight, or, in fancy terminology, a geodesic. Another way
of saying this is that each astronaut is continuously experiencing apparent weightlessness. This is the equivalence principle at
But after some time has passed, they notice that they are no
longer receding from one another.
Ah, Two, this is One, over.
Roger, One. Hey, have you done some kind of course correction? Youre not moving away from me anymore.
No, Two. Im going straight. But you must have fired your
jets, because youve stopped moving away from me.
The situation is similar to the one with the girls in Mumbai
and Los Angeles on p. 109. The equivalence principle is local, so



locally, the effects of gravity can be made to vanish by adopting

an inertial (free-falling) frame of reference. But globally, that
doesnt work. There is a planet in between the astronauts, and
even though its surrounding gravitational field can be considered
to be an illusion locally, it is inescapably real globally. It will have
real, observable effects, because each astronaut ends up falling
back toward the planet and crashing into the surface.
This is what we mean by curvature of spacetime. We can
have two straight lines in spacetime that are initially diverging,
but that later reconverge. Just as the bugs are forced to conclude that their world is curved, the astronauts will attribute
their unfortunate fates to the fact that their spacetime is curved.
The curvature is being caused by the gravity of the planet. In
relativity, this is how we describe gravity: gravity is curvature of


Deflection of starlight

Since the mass of a ray of light is zero, Newton would have expected it to be immune to gravity. But in general relativity,
several different arguments lead us to expect that a ray of light
can be deflected by a gravitational field:
1. A ray of light does have energy, and energy is equivalent to
2. Suppose, for example, that we send a ray of light in the
horizontal direction here on earth. In a free-falling frame,
gravity doesnt exist, and by symmetry there is no reason
for the ray to bend in any direction. But if the ray is
straight in a free-falling frame, then it is curved in a frame
fixed to the earths surface.
3. Consider the following analogy with special relativity. Suppose that time dilation affected some clocks but not others,
and length contraction affected some rulers but not others. Then we would not be able to interpret the Lorentz



transformation as describing a distortion of spacetime itself. Similarly, suppose that gravity affected the motion of
some objects (rocks) but not others (flashlight beams). We
would then have good reason to abandon general relativity,
which describes gravity as a curvature of spacetime itself.

The first important experimental confirmation of relativity

came in 1919 when stars next to the sun during a solar eclipse
were observed to have shifted a little from their ordinary position,
as shown in the photograph below. (If there was no eclipse, the
glare of the sun would prevent the stars from being observed.)
Starlight had been deflected by the suns gravity. The picture
is a photographic negative, so the circle that appears bright is
actually the dark face of the moon, and the dark area is really the
bright corona of the sun. The stars, marked by lines above and
below them, appeared at positions slightly different than their
normal ones.



The diagram below shows an explanation of the effect. A ray

of light from the star arrives from a different direction than the
one from which it would have come if the suns gravity had not
interfered. This leads to a change in the stars apparent position
in the sky.


No extra dimensions

In the parable of the bugs (section 9.1, p. 118), we visualized

their universe as a two-dimensional surface embedded in a threedimensional space. The third dimension just happened to be
invisible to the bugs. In science, we usually try not to invoke the
existence of things that cant be empirically observed, such as
invisible pink unicorns (cf. section 4.1, p. 63). To the bugs, there
are no extra dimensions. They exist in two dimensions of space
and one dimension of time, abbreviated as 2 + 1. Similarly, you
and I live in 3 + 1 dimensions. Although we may find it useful,
when we want to conceptualize curvature, to draw pictures like
the pictures of the bugs on a sphere, that doesnt mean that
our universe actually has extra dimensions that are invisible to
us. Nobody has ever succeeded in observing a fifth dimension,
although attempts have been made using particle accelerators
(Franceschini et al., arxiv.org/abs/1101.4919).




Parallel transport

This unobservability of extra dimensions has some interesting

consequences. In figure 1, a woman walks from the earths equator to the north pole, and as she does so, her body rotates by
90 degrees relative to the stars. If she was living in the bugs
universe, the stars wouldnt exist. But she isnt, they do, and
she can detect the rotation, e.g., when she was at the equator,
the pole star was on the horizon, but once she got to the north
pole, it was directly overhead.

Even if the weather was always cloudy, so that she could never
observe any external reference points, it would still be possible
for her to verify, using only local measurements, that she had
rotated. For example, as in figure 2, she could carry a gyroscope
with her.3 When she starts out at the equator, the gyroscopes
axle is perpendicular to her body. When she gets to the north
pole, its parallel to her spine. Since the gyroscope is supposed
to maintain its orientation, she can infer that it was her body
that rotated.
But consider how this would work out for the bug in figure 3.
We can imagine that the bug owns a smartphone with a built-in
gyroscope. But because there is no third spatial dimension in the
bugs reality, the smartphones screen always lies in the plane of
the surface, and if it displays a fixed direction as an arrow on the
phones screen, that arrow always lies in that plane as well. If
The figure shows a toy mechanical gyroscope, which wouldnt actually be
practical for this purpose, but many smart phones contain a tiny gyroscope
fabricated as part of a silicon chip.



the bug travels 90 degrees around the sphere, its not possible to
obtain the same result as in figure 2, because that would require
the arrow to point in a direction that doesnt exist. The bug
simply sees the arrow point steadily in the direction that the bug
is going. We say that the arrow has been parallel-transported
along the surface of the sphere. Parallel transport means taking
a vector and moving it to a new location along a certain path,
without allowing any external influence to twist the vector (or
change its magnitude).
Similarly, suppose that the bug had a frictionless hockey puck.
He could push the puck forward, and it would glide with a certain
momentum vector. No external influence is deflecting the puck,
so its momentum vector is parallel-transported.

Figure 4 duplicates figure 3: our bug parallel-transports his

cell-phone gyroscope from what we would describe as his worlds
equator up to its north pole. Meanwhile, in figure 5, his
friend starts from the same point and ends up at the same destination, but via a different path: he first travels west 90 degrees
along the equator, then goes north to the pole. When the bugs
start out, they make sure that their gyroscopes agree. When
they are reunited, they find that the gyroscopes are out of whack.
Each bug could claim that the others gyro is broken, but in reality what has happened is an effect of the curvature of their
world. In fact, this is a good general definition of what we mean
by curvature: when curvature is present, parallel-transport from
point A to point B can give a result that depends on the path
This path-dependence of parallel transport was verified di-



rectly in 2005 in a satellite-based experiment called Gravity Probe

B. On board the satellite was a gyroscope in the form of a silicon
ball spinning frictionlessly in a vacuum inside a box. The satellite was orbiting the earth, so on a spacetime diagram we would
visualize its world-line as a helix. Parallel-transporting the gyroscope along this helical world-line gives a different result than
if its path through spacetime had not been orbiting. If the gyroscope had then been brought down from orbit and compared with
one left on the surface of the earth, they would have disagreed
by a small amount, due to the earths gravity: the earths gravitational field is a curvature of the surrounding space. In reality,
the gyroscope was too delicate to be handled in this way without
disrupting its motion, so the effect was observed as a change in
the orientation of its axis relative to the stars. Over the one-year
duration of the mission, the axis twisted around by an angle of
about 0.002 degrees, in good agreement with the predictions of
general relativity.

Chapter 10

In Newtons nonrelativistic description (ch. 7, p. 93), gravity was
a universal attraction between masses and other masses. It was
a force, and it acted instantaneously from a distance. The relativistic description is completely different. Matter, such as the
planet in the figure on p. 122, causes curvature in the spacetime
around it. This curvature then influences the motion of matter.
The relativistic picture is called Einsteins theory of general relativity. The word general is used because in special relativity,
the curvature of spacetime is assumed to be zero; general relativity generalizes the theory to allow the curvature to be nonzero.
Relativist John Wheeler summarizes this in a slogan:
Spacetime tells matter how to move; matter tells spacetime how to curve.
Section 13.1 further fleshes this out.
Wheelers description requires us to define what we mean by
matter. Newton would have said that mass was the correct way
to measure the quantity of matter. But for the reasons discussed
in section 9.4, matter in general relativity should be taken to
mean mass-energy, not just mass.
One of the most surprising predictions of general relativity
is the existence of black holes, which are formed when matter
is sufficiently compressed by gravity. To understand how this



can happen, we need to understand how matter should be expected to behave under extreme conditions. Luckily this turns
out to be unexpectedly easy. Understanding the detailed properties of matter is the highly technical province of specialists such
as metallurgists, physical chemists, and condensed-matter physicists. But, surprisingly, relativity itself makes some more general
predictions about matter that are independent of its detailed


No nongravitating matter

One such prediction is that there is no such thing as matter that

is immune to gravity; anything that has mass-energy is affected
by gravity. We see objects in everyday life that superficially
seem to have such an immunity. Birds can fly, but thats not
because they ignore gravity. Birds exploit forces from the air
acting on their wings. The same kind of explanation holds for a
speck of dust floating in the air, the microscopic drops of water
that form clouds, helium balloons, or air molecules themselves.
In particular, if we use a vacuum pump to suck the air out of a
flask, the flask becomes lighter, because the air that was removed
had weight. If air didnt have weight, the earths gravity wouldnt
keep the planets thin film of atmosphere from escaping, and our
world would be an airless body like the moon. This universal
susceptibility to gravity even extends to things that nobody but
a relativist would call matter, such as the rays of sunlight in the
1919 test of general relativity described on p. 124.
Suppose, on the contrary, that we could find some form of
exotic matter call it FloatyStuffTM that had the ordinary
amount of inertia, but was completely unaffected by gravity. If
FloatyStuff existed, it would cause us to doubt the equivalence
principle. The basis for the equivalence principle is the universality of free fall, which is violated by FloatyStuff.
To dramatize this a little more, say that alien gangsters land
in a flying saucer, kidnap you out of your back yard, konk you



on the head, and take you away. When you regain consciousness, youre locked up in a sealed cabin in their spaceship. You
pull your keychain out of your pocket and release it, and you
observe that it accelerates toward the floor with an acceleration
that seems quite a bit slower than what youre used to on earth,
perhaps a third of a gee. There are two possible explanations for
this. One is that the aliens have taken you to some other planet,
maybe Mars, where the strength of gravity is a third of what
we have on earth. The other is that your keychain didnt really
accelerate at all: youre still inside the flying saucer, which is in
interplanetary space and accelerating at a third of a gee, so that
it was really the deck that accelerated up and hit the keys.
There is absolutely no way to tell which of these two scenarios is actually the case unless you happen to have a chunk of
FloatyStuff in your other pocket. If you release the FloatyStuff
and it hovers above the deck, then youre on another planet
and experiencing genuine gravity; your keychain responded to
the gravity, but the FloatyStuff didnt. But if you release the
FloatyStuff and see it hit the deck, then the flying saucer is accelerating through outer space. In this situation, you are able to
tell the difference between a gravitational field and an accelerated
frame of reference, which according to the equivalence principle
are supposed to be equivalent things.
Thus the existence of matter that is immune to gravity is forbidden by the equivalence principle. It is an ironclad prediction
of general relativity that no such form of matter exists. If any
such material were ever discovered, general relativity would be
For similar reasons, we can never shield against gravitational
fields in the same way that we can with electrical, magnetic, or
other forces. If we could, then we could take any ordinary piece of
matter, such as a brick, and turn it into a gravity-immune object
by wrapping it in gravitational shielding. This is why, in our
operational definition of an inertial frame of reference (section
8.2, p. 105), we shielded against all forces except for gravity,
which would have been impossible.



As an example, the figure shows an old ad that was frequently

found in comic books. If the ad claimed that the car worked by
antigravity, then we could be pretty sure that it was a scam,
since it would violate general relativity. But a careful reading
of the text shows that the car is supposed to work by floating
on a cushion of air, which is more scientifically plausible, and
in fact, people who bought the car way back when have posted
descriptions on the internet stating that the car did work. (But
note that its only 11 inches long, so you cant ride in it.)
Ive structured this book so that the next few chapters revolve
around black holes. As well see in more detail later, there are
some pop-culture ideas about black holes that turn out to be
right, and others that are wrong. One of the correct ones is that
once youve fallen into a black hole, you can never escape. The
nonexistence of gravitationless matter and gravitational shielding
help to explain why this should be so. If you had a flying car
that was made out of gravitationless matter, it certainly does
seem like it would be able to escape a black holes gravity.


No repulsive gravity

The black holes that we observe are believed to have formed by

the runaway gravitational collapse of stars or clouds of gas and
dust. This would be much like the formation of objects such as
the earth and the sun, except that the collapse would have progressed much further, to extremely high densities. When the idea
of black holes was first proposed (long before they had actually



been observed), various objections were raised to their formation,

one of them being that we dont necessarily know the properties
of matter under such extreme conditions. By analogy, suppose
that we place two hydrogen atoms at some small distance from
each other, say a distance equal to a few of their own diameters. They will experience an attractive electrical force and will
fall toward each other. But once they get close enough, so that
their electron clouds start to overlap significantly, the attraction
turns into a repulsion. The collapse stops, and we end up with
a stable hydrogen molecule, the form of the gas that is actually
encountered normally.
Is it possible, then, that under certain conditions gravity
might also behave repulsively, possibly preventing the formation
of black holes? A silly pop-culture version of this, from the Rocky
and Bullwinkle cartoon series, is the fictional substance upsidaisium, which falls up rather than down. The figure shows a soldier
guarding a military stockpile of upsidaisium bricks, which have
to be kept staked down so they dont fly up into the sky.

If upsidaisium were discovered, it would violate the universality of free fall, and therefore invalidate the equivalence principle
and disprove general relativity.



But there is a way to get repulsive gravity without throwing

general relativity down the toilet. Recall that mass is defined not
in terms of weight but as a measure of inertia (section 5.1, p. 71).
Suppose that we have a brick whose mass is minus one kilogram.
This is admittedly very strange. For example, if I hit such a brick
with a baseball bat toward the east, it would accelerate to the
west. If I whacked a car with it, the bricks momentum would
be in the opposite of the expected direction, so the sheet metal
would pucker outward rather than forming an inward dent. But
if this brick also had negative weight (like upsidaisium), it would
fall down, just like normal matter: since it responds to forces in
the opposite of the usual way, the upward force of gravity would
accelerate it downward. Thus we could have repulsive gravity
while preserving the equivalence principle.
The nonexistence of repulsive gravity, or negative mass-energy,
is therefore not completely ruled out by general relativity, and we
have to posit it as an extra assumption. Giving an exact mathematical statement of this assumption is complicated and beyond
the scope of this book, but the foregoing discussion should make
it clear what the general idea is. It turns out that there is more
than one way of giving an exact formulation, and the different
ways are not equivalent to one another; some are stricter requirements and some are weaker. These are referred to as energy
conditions. If we look at the ordinary forms of matter that surround us in our solar system, or that we have been able to test
in the laboratory, all of them obey all of the energy conditions.
A notable exception is dark energy, a recently discovered, mysterious phenomenon that seems to pervade the universe, but that
has a negligible effect except at cosmic scales.


Finite strength of materials

A final possible objection to the possibility of runaway gravitational collapse is that with many materials, the more we compress
them, the more they resist compression. For example, pumping



up the tire of a bicycle is relatively easy at first, but as the pressure inside the tire gets greater and greater, it gets harder and
harder to pump. Now suppose we had a powerful mechanical
pump. It could compress air to much greater pressures than we
could achieve using a hand pump, but at a certain point, something new would happen: under sufficient pressure, air turns
from a gas into a liquid. (A similar example is the carbon dioxide canisters hidden behind the soda dispensers at restaurants.
The CO2 inside has been pressurized so much that it is a liquid.)
A liquid has qualitatively different properties than a gas, and in
particular it strongly resists further compression. This is why,
when you squeeze a water balloon, it just bulges outward in the
gaps between your fingers.
Under higher and higher pressures, matter is believed to pass
through a series of abrupt changes similar to the gas-liquid transition more about this in section 11.1, p. 137. With each such
transition, it crunches down into a denser and yet more exotic
state. Our scientific knowledge is always limited, and these are
conditions that we cant replicate in the laboratory. Is it possible that beyond the states of matter currently hypothesized there
is some other, as yet unsuspected, form of matter that is even
more dense, and that represents the ultimate possible state of
compression of matter?
Surprisingly, relativity allows us to answer this question, and
the answer is no. It tells us that there must be a limit to the
ability of matter to resist compression.
The key here is to consider the speed at which vibrations
travel in a material. When you pop a champagne bottle, the
disturbance in the air spreads outward as an audible sound. The
speed of sound in air is fast several hundred meters per second.
In a more incompressible substance like water, the speed of sound
is about five times greater. This is because stronger forces now
resist compression, and these stronger forces cause the water to
accelerate more violently. If a substance had infinite resistance
to compression, then it would be perfectly rigid, and the speed of
sound in that substance would be infinite. But relativity tells us



that this is impossible. No signal or physical effect can propagate

at a speed greater than c.

We therefore conclude that there is some limit to the ability

of matter to withstand compression. Every substance, like the
crushed cars in the photo, will compress if squished hard enough.
Thus it seems that if a sufficiently massive body collapses gravitationally, the result must be a runaway collapse.

Chapter 11

Black holes


Collapse to a singularity

Lets talk about death. Our sun is very close to the end of the
time during which it will be the right brightness to support life on
our planet. As it nears total depletion of the hydrogen fuel that it
burns, it will pass through a series of complicated changes, at first
subtle and gradual but then more rapid and violent. Astronomers
know about these changes because other stars, further down the
road, provide snapshots of the process. Near the beginning of
the suns process of dying, our little planet will be completely
sterilized, and later it may be destroyed outright.



Once the fuel gives out, our sun loses its source of heat. Just
as the heat inside a hot air balloon keeps it inflated, the heat
energy produced in our sun is currently what prevents it from
crumpling. With the heat turned off, there will be nothing to
oppose gravity, and the sun will begin to collapse gravitationally.

But this will not, according to our current understanding, be

a runaway collapse. Our sun is after all a pretty small and insignificant star, and its gravity is not exceptionally strong. In the
end, we think matters resistance to compression will be strong
enough to slow the collapse to a stop, leaving our dead star in a
very dense form called a white dwarf. A new and final equilibrium
will have been achieved. White dwarf matter has the fantastic
density of about a ton per cubic centimeter, far beyond anything
we can achieve in the laboratory.1 But this is not pure theoretical
extrapolation. Our galaxy is littered with white dwarfs, which
are the most common type of stellar corpse.
In 1930, at the age of 19, the astrophysicist Subrahmanyan
Chandrasekhar (1910-1995) demonstrated in a theoretical calculation that white dwarf matter could not stand up to more than

The white dwarf in the figure is not drawn to scale; it would actually be
about the size of the earth, which would be too small to see on the page.



a certain amount of pressure. Inside a big enough dying star, the

atoms themselves begin to collapse under the pressure of gravity,
with the electrons and protons fusing to form neutrons. The result is a ball made mostly of pure neutrons, called a neutron star.
Observational evidence for the existence of neutron stars came
in 1967 with the detection by Bell and Hewish at Cambridge
of a mysterious radio signal, repeating at an interval of exactly
1.3373011 seconds. The signals observability was synchronized
with the rotation of the earth relative to the stars, rather than
with legal clock time or the earths rotation relative to the sun.
This led to the conclusion that its origin was in space rather than
on earth, and Bell and Hewish originally dubbed it LGM-1 for
little green men. The discovery of a second signal, from a different direction in the sky, convinced them that it was not actually
an artificial signal being generated by aliens. Bell published the
observation as an appendix to her PhD thesis, and it was soon
interpreted as a signal from a neutron star. Neutron stars can be
highly magnetized, and because of this magnetization they may
emit a directional beam of electromagnetic radiation that sweeps
across the sky once per rotation the lighthouse effect. If
the earth lies in the plane of the beam, a repeating signal can be
detected, and the star is referred to as a pulsar.
As we saw in section 10.3, p. 134, relativity predicts that we
cant go on discovering more and more compressed forms of matter. (Its a bit like the situation of Columbus and Magellan: they
may have been surprised to come across two or three new continents, but they knew that the world was round, so at some point
there had to be an end to what there was to discover.) For a sufficiently massive star, we expect that gravity will win and matter
will lose. There will be a runaway collapse. The end result is expected to be a black hole: a star with all its mass squeezed down
to a space smaller than an atom of ordinary matter. This tiny
object, which is a mathematical point according to general relativity, is called a singularity. It represents a point where either
the laws of physics break down completely, or general relativity
stops making sense and should not be trusted.




The event horizon

The most distinctive property of black holes is that they are

expected to be black. To see why, consider what we know about
gravitational time dilation (section 8.5, p. 110).
If we compare the flow of time at point A on the earths
surface with its rate of flow at some distant point B such as a
GPS satellite, we get a difference, but its very small, only a part
per billion perhaps. We can make the effect a little bigger by
choosing B to be farther away, but we rapidly reach a point of
diminishing returns, because the earths gravity dies out rapidly
with distance. We could also try making point A underground,
but again this doesnt produce any vastly increased effect. The
reason is the same: the earths gravity gets weaker as you go deep
inside it. If this seems surprising, consider what you would feel
if you could go to the point exactly at the center of the earth:
the attraction of the earths mass would pull on you equally in
all directions, so there would be no gravity at all. So we can
maximize the gravitational time dilation of A relative to B by
putting A at the earths center and B infinitely far away, but the
result will still be in the parts-per-billion range. The ultimate
reason for this is simply that the earth has a certain size, and we
cant get any closer to all of its mass than that.
With a white dwarf we can do quite a bit better. The mass
is a million times greater, so we can get a time dilation effect
almost in the parts-per-thousand range. The detection of this
effect was one of the early experimental tests of relativity.
What about a neutron star? Whereas a white dwarf is about
the size of the earth, a neutron star is the size of a city. The mass
isnt much more, but we can get closer to all of it. The result is
time dilation of something like ten percent.
Now you can see where all of this is leading. Because a black
hole is essentially a pointlike object, there is no limit on how
close you can get to all of its mass if you dare. Although the
math is too complex to go into here, it turns out that if you get
within about 30 kilometers of a ten-solar-mass black hole, the



time dilation becomes infinite.

Now lets think about what that means. We have an imaginary sphere centered on our black holes singularity, with a radius
of 30 km. The surface of this sphere is called the event horizon.
Suppose that you fly down to the event horizon in your rocket
ship and then try to transmit a radio message back home. You
speak the first word of your message, and then the second word.
To you, the time between these words is a fraction of a second.
But to a distant observer receiving this message, the time interval between the words equals eternity. In other words, it is not
possible to transmit information to a distant point from the event
horizon or anywhere inside it.
Similarly, suppose that what is emitted from the event horizon isnt an information-bearing signal but simply a light wave.
At emission, the time between one vibration of the wave and the
next is very short, but to a distant observer its an infinite time.
Clearly the light wave simply isnt escaping.





Because light cant escape from inside the event horizon, we can
never hope to observe black holes by seeing them shine, as normal
stars do. There are nevertheless several methods that have either
been used already or may be used in the future for detecting black
One method, which is so far purely hypothetical, is gravitational lensing, as shown in figures 1 and 2. Figure 1 shows
a telescopes magnified view of a distant galaxy. Now suppose
that, by pure luck, a black hole happens to lie very nearly along
our line of sight to this galaxy. As in the 1919 test of general relativity described on p. 124, any rays of light that happen to pass
very close to the black hole (but without entering its event horizon) will be bent by the black holes gravity. The result would
be a type of distortion of the image, as in figure 2. Although
gravitational lensing has been detected many times, lensing by a
black hole has not yet been observed.

A far easier method is to exploit the fact that, as in the

simulated view in figure 3, black holes are often actively feeding
on nearby matter. In this simulation, an entire star has wandered
too close to the black hole, which is at the position marked with
crosshairs. The star has been broken up by tidal forces (section
7.4, p. 100) and is now falling toward the black hole as a cloud
of gas. Internal friction causes the gas to become so hot that it
glows in the x-ray part of the spectrum. These x-rays are a highly
distinctive sign of a black hole. The object known as Cygnus X1 is the best-studied example. This X-ray-emitting object was



discovered by a rocket-based experiment in 1964.

Around the turn of the 21st century, new evidence was found
for the prevalence of supermassive black holes near the centers of
nearly all galaxies, including our own. Near our galaxys center
is an object called Sagittarius A*, detected because nearby stars
orbit around it. The orbital data show that Sagittarius A* has a
mass of about four million solar masses, confined within a sphere
with a radius less than 22 million kilometers. There is no known
astrophysical model that could prevent the collapse of such a
compact object into a black hole, nor is there any plausible model
that would allow this much mass to exist in equilibrium in such
a small space, without emitting enough light to be observable.
The existence of supermassive black holes is surprising. Gas
clouds with masses greater than about 100 solar masses cannot
form stable stars, so supermassive black holes cannot be the endpoint of the evolution of heavy stars. Mergers of multiple stars
to form more massive objects are generally statistically unlikely,
since a star is such a small target in relation to the distance
between the stars. Once astronomers were confronted with the
empirical fact of these supermassive black holes existence, a variety of mechanisms was proposed for their formation. Little is
known about which of these mechanisms is correct.


Light cones

A typical scheme for evading the one-way nature of the event

horizon would be to put a space probe on the end of a rope,
lower the rope down through the horizon, and then pull it back
out. The plan wont work, because tidal forces will break the
rope. Why cant we just use a stronger rope? The answer is that,
as weve seen in section 10.3, p. 134, relativity imposes ultimate
limits on the strengths of materials. This example demonstrates
a more general fact, which is that the event horizon is not just a
barrier that is difficult to escape through. The reason its called
the event horizon is that it marks a fundamental disconnection



of cause and effect. Events inside the horizon can never cause
events outside.
The figure on the facing page shows a good way to visualize this. Its a standard spacetime diagram, with the time axis
running vertical and space horizontal. The black hole is stationary in this frame of reference, so the singularitys world-line is a
vertical line, and so is the event horizons. Astronauts Alice and
Bob start at rest, at unequal distances from the black hole, and
under the influence of gravity they begin to accelerate toward it
in free fall.
Since theyre so close to a black hole, you might imagine that
they would feel very heavy, but theyre free-falling, so they both
experience apparent weightlessness (section 8.3, p. 107). This is
the equivalence principle at work. Locally, gravity is an illusion
that we only experience if we adopt a frame of reference thats
not inertial (not free-falling).
Suppose Alice carries out physics experiments in the small
region of spacetime indicated by the dashed square. She will find
absolutely nothing unusual, and everything will be explainable in
terms of special relativity, because the curvature of spacetime is
not detectable on such a small scale. (Similarly, we dont worry
about the curvature of the earth when driving around town.) In
particular, she will be able to define a future light cone for herself,
as special relativity says she can always do.
But now consider the situation shown near the top of the diagram, when Alice has crossed the event horizon. We know that
if she emits a ray of light from this position, aiming it outward, it
will not escape the black hole. But to Alice, in her frame of reference, special relativity is still locally valid, and the ray of light
travels, as light always does, at c. How can this be, if the light
isnt going to escape? The resolution of the paradox is that if we
draw her light-cones on this type of diagram, theyre tipped. In
her local frame of reference, light moves at the expected speed in
both the inward and outward directions, and if she were drawing
the light cone, shed draw it upright. But to us, in our global
view, its tipped.





The light cone, as weve already seen, is not fundamentally

about light. Its about mapping out the possible cause-and-effect
relationships between events. As Alice crosses the event horizon,
her entire future light cone tips over so far that it all points
inward, not outward. Alice can no longer physically escape, nor
can she send a signal back to Bob. For her, there is no future
other than the region inside the black hole.


Penrose diagrams

The figure with the tipping light cones on p. 145 is one way of
visualizing the curved spacetime around a black hole. In this
section I introduce a different method called a Penrose diagram,
named after mathematician and relativist Roger Penrose, which
is in some ways nicer. It starts from the simple idea of a vanishing
point, demonstrated in the image below of the view through a
cars windshield.

The parallel lines painted on the asphalt appear to converge

toward a vanishing point in the distance. This is simply because
distant objects look smaller, which is a type of distortion, but its
a distortion that were accustomed to and dont normally think
about. In fact, its a useful distortion. The ancient Egyptians
didnt understand how to handle perspective in their art, so they
couldnt render a landscape the way its shown in this figure,



simulating the eyes view. Without the useful distortion of perspective, we wouldnt be able to simultaneously perceive both
the distant mountains and a nearby feature like a bush. If the
bush were rendered on the same scale as the mountain, we would
either have to make the bush microscopic, and too small to see,
or wed have to make the mountains so big that they wouldnt fit
in the picture. By the way, the road also has a vanishing point
behind your head, which is shown in the rear-view mirror. You
dont normally see this one, but if you had eyes in back as well
as in front, like some jumping spiders, it would look the same to
you as the one in front.
In a Penrose diagram, shown in figure 1 below, we have the
same idea, but this is now a spacetime diagram, so the vanishing
point at the top doesnt represent a point infinitely far in space
but one infinitely far in the future. As in perspective art, we
distort scales in order to fit everything in. Even though the
vanishing point is infinitely far in the future, we draw it just
a few centimeters away from the bottom of the diagram. (The
distance scale also gets more and more distorted as we move out
toward the corners on the sides. These corners represent infinitely
distant space at the present time.) The two thinner lines in the
diagram are analogous to the painted white lines in the photo of
the road. They are parallel, i.e., they always maintain the same
distance away from one another in space.

Keep in mind that all weve drawn so far is a Penrose diagram

for flat spacetime, i.e., special relativity, with no gravity. There



are no black holes lurking in our landscape. All of the distortions are distortions in the way weve draw the diagram, not real
curvature of spacetime.
Figure 2 shows some light cones superimposed on the Penrose
diagram. Despite all the distortions, the light cones dont tip.
This is the big advantage of a Penrose diagram. Light cones
spell out the logic of possible cause and effect relationships, so
theyre very important. When we refrain from tipping them, we
make it easier to look at the diagram and reason about the light
Figures 1 and 2 started from the present moment in time and
went forward into the future. But we can also represent the past
on the same Penrose diagram, as in figure 3, making the result
into a diamond. This is like adding in the rear-view mirror in
the original photo of the landscape.
So far this has all been in flat spacetime. Before we go
into curved spacetime, lets warm up by thinking about ordinary
curved surfaces, like the surface of the earth. Because the world
is round, we can never reproduce it perfectly on a flat map. To
flatten its curved surface, we must distort, fold, or cut it. The figure below shows two of the many possible ways of accomplishing

In the version on the left, the latitude-longitude grid is rendered such that where the lines intersect, they always form right
angles. We pay for this feature with an extreme distortion of
lengths and areas. For example, Greenland looks almost as big
as Africa, whereas in reality Africa is much larger.



The version on the right has its own pros and cons. Lengths
and areas are much more accurately represented, but the latitudelongitude boxes have funny nonrectangular shapes.
Something exactly analogous happens when we represent curved spacetime on a flat diagram. Two options are shown in the
diagram below. The one on the right is just like the one we drew
in section 11.4, with the light cones tipping over drunkenly. This
tipping is analogous to the non-perpendicularity of the lines of
latitude and longitude in the right-hand map of the earth. As
in the earth map, we have the advantage that the sizes of things
make sense: the interior region of the black hole is of finite size,
whereas the exterior region is much bigger limited in size only
by the edge of the page.

The figure on the left is a Penrose diagram for a black hole.

Here the light cones are all lined up in an orderly way, which is
analogous to the orderly rectangular structure of the grid lines
in the left-hand map of the world. This is an overwhelming advantage, since the light cone dictates all of the possible cause-



and-effect relationships. For this reason, professional relativists

use Penrose diagrams much more frequently than the ones with
tipping light cones. The disadvantage of the Penrose diagram is
that sizes are not represented accurately. For example, the triangle representing the interior of the black hole looks half as big
as the diamond showing the whole outside universe.


Tests of general relativity in weaker


Figure 1 on the facing page contrasts the predictions of Newtons

laws and general relativity for an object whose orbit brings it inside the event horizon of a black hole. They are qualitatively
different. According to Newton, the event horizon doesnt even
exist, so nothing special happens. The objects orbit is a Keplerian ellipse. But according to relativity, the object can never
come back out once it has passed the event horizon. It impacts
the singularity at the center.
Relativity also predicts smaller deviations from Newtonian
behavior under less extreme conditions. Figure 2 shows that for
a body in an elliptical orbit that only comes moderately close to
the black hole, the main relativistic effect is a gradual twisting
of the ellipses orientation. This is called precession. By the way,
these diagrams were traced from a cool browser-based simulation,
which you can play with at m4r35n357.github.io/orbits/.
If we are to observe such an effect in reality, we want to look
for the strongest possible gravitational field. This makes it logical
to examine the orbit of the planet Mercury. Mercurys orbit is
significantly elliptical, and just as importantly it comes closer to
the sun than any other planet, therefore experiencing stronger
relativistic effects.
Before Einstein published the general theory of relativity, it
had already been observed that Mercurys orbit precessed very
slightly, by 1.6 degrees per century. Nearly all of this effect had
been explained by Newtonian effects such as the gravitational



influence of Jupiter and the slightly nonspherical shape of the

sun. But after all of these had been accounted for, there remained
an unexplained 0.012 degrees per century. One of the first things
Einstein did after finding the field equation of general relativity
was to calculate this anomalous precession. It came out right,
and he later recalled, I was beside myself for several days with
joyous excitement.


Speed of light

The equivalence principle tells us that locally, gravity is an illusion. To any local, free-falling observer, the laws of physics and
the properties of spacetime seem absolutely normal. The laws
of physics tell us that the speed of light is a universal constant,
and an observer sees nothing abnormal in local experiments that
measure the speed of light.
But the equivalence principle is only a local thing, so globally
things look different. Recall that on a spacetime diagram, the
speed of an object is measured by how much its world-line tilts
away from vertical, and the speed of light is indicated by a 45degree tilt. In the diagram in section 11.4, the fact that the light
cones are tipped make it look as though light isnt moving at the
speed of light! In fact you are free to tilt your copy of the book to
any angle you like, and by doing this you can view any particular
light cone the way an observer in that local area would prefer to
draw it upright and law-abiding.
The basic problem here is that when we try to visualize
everything-that-is all-at-once, we run up against the hard realities of relativity. An observer O has no technique of measurement
that can tell her unambiguously the state of motion of some distant thing T. Such a measurement technique would require sending signals from T to O, and these signals could be distorted
by effects such as Doppler shifts. Furthermore, we would have to
ask motion relative to what? The local points of reference near
T are different than the ones near O. A good example of such





an ambiguity is the ambiguity of parallel transport (section 9.6,

p. 126). In any local region of spacetime, we can define velocity
vectors for things like basketballs and rays of light. But if we
want to ask how a distant observer would perceive such a velocity vector, we have no unambiguous way to answer the question.
This would require parallel-transporting the velocity vector over
to where the observer is sitting, but parallel-transport is ambiguous and path-dependent.



Chapter 12


No action at a distance

The Newtonian picture of the universe has particles interacting

with each other by exerting forces from a distance, and these
forces are imagined to occur without any time delay. But as
weve seen, relativity forbids this. The universal speed c is the
maximum speed at which cause and effect can propagate. This
blanket prohibition applies to any force at all. In Star Wars, Obi
Wan Kenobi senses the destruction of the planet Alderaan and
says, I felt a great disturbance in the Force, as if millions of
voices suddenly cried out in terror, and were suddenly silenced.
No matter what kind of force the Force is, this is impossible.
The planet is light-years away, so any information about its destruction can only arrive after a time delay of years.



Since forces cant be transmitted instantaneously, it becomes natural to imagine force-effects spreading outward from their source
like ripples on a pond, and we then have no choice but to impute some physical reality to these ripples. We call them fields,
and they have their own independent existence. Gravity is trans155



mitted through the gravitational field. Besides gravity, there are

other fundamental fields of force such as electricity and magnetism. Ripples of the electricity and magnetism are light waves.
Even empty space, then, is not perfectly featureless. It has
measurable properties. This concept made a deep impression on
Einstein as a child. He recalled that as a five-year-old, the gift
of a magnetic compass convinced him that there was something
behind things, something deeply hidden.
The smoking-gun argument for this strange notion of traveling force ripples comes from the fact that they carry energy. In
figure 1, Alice and Betty hold magnets A and B at some distance
from one another. If Alice chooses to move her magnet father
from Bettys, as in figure 2, Alice will have to expend some energy to fight against the magnetic force, burning off some of the
calories from that chocolate cheesecake she had at lunch.

But now suppose, as shown in figure 3, that Betty decides to

play a trick on Alice by tossing magnet B far away just as Alice
is getting ready to move magnet A. We have already established
that Alice cant feel magnet Bs motion instantaneously, so the
magnetic forces must actually be propagated by a magnetic field.

12.2. FIELDS


Of course this experiment is utterly impractical, but suppose for

the sake of argument that the time it takes the change in the
magnetic field to propagate across the diagram is long enough
so that Alice can complete her motion before she feels the effect
of Bs disappearance. She is still getting stale information about
Bs position. As she moves A to the left, she feels resistance,
because the field in her region of space is still the field caused by
B in its old position. She has burned some chocolate cheesecake
calories, and it appears that conservation of energy has been
violated, because these calories cant be properly accounted for
by any interaction with B, which is long gone.
If we hope to preserve the law of conservation of energy, then
the only possible conclusion is that the magnetic field itself carries
away the cheesecake energy. In fact, this example represents
an impractical method of transmitting radio waves. Alice burns
calories by moving magnet A, and that energy goes into the radio
waves. Even if B had never existed, the radio waves would still
have carried energy, and Alice would still have had to do work in
order to create them.
Although I referred to magnets in this argument, I never appealed to any properties of the magnetic force, so it follows that
force-ripples in any field carry energy. For example, suppose that
super-powerful aliens, angered by the advent of disco music, come
to our solar system on a mission to cleanse the universe of our
aesthetic contamination. They apply a force to our sun, causing
it to go flying out of the solar system at a gazillion miles an hour.
Newton expects the gravitational force of the sun on the earth
to immediately start dropping off, and and since sunlight takes
eight minutes to get from the sun to the earth, the change in
gravitational force would be the first way in which earthlings
learn the bad news the sun would not visibly start receding
until a little later.
Newton is wrong. Both effects would travel at c, and they
would be detected simultaneously on earth.




Electromagnetic waves

And God said, Let there be light: and there was light. Light
is pretty important to us, as implied by this passages prominent placement in the creation myth. Light is a vibration of
the electric and magnetic fields, and to make sense out of that
statement, you need to have at least some idea of what electric
and magnetic fields are. The operationalist philosophy (section
4.1, p. 63) is that this means knowing how to measure them. A
magnetic compass measures the direction of the magnetic field.
A crude method of measuring an electric field is to rub a balloon
against your hair in order to put some electric charge on it, and
then release the balloon; any ambient electric field will cause the
balloon to accelerate in a certain direction, which is the direction
of the field.
Although physicists didnt realize it until late in the game,
it turns out that electric and magnetic fields are very closely
related. If an observer in frame of reference A detects only an
electric field, an observer in frame B, moving relative to A, sees a
mixture of magnetic and electric fields. Similarly, a field that is
purely magnetic in one frame is a mixture in another frame. For
related reasons, a changing electric field creates or induces
a magnetic field, and vice versa. This is the principle behind
devices such as generators and transformers.

Vibrations in these fields will be emitted by any electric charge

that shakes back and forth. For example, in a candle flame,



the high temperature indicates rapid vibration of the atoms and

charged subatomic particles in the flames. When you transmit
radio waves from the radio antenna hidden inside your cell phone,
electrons are wiggling back and forth between the ends of a hidden antenna.
Because of the close relationship between the two fields, any
vibration of the electric field produces a vibration of the magnetic field, and vice versa. Therefore there are no purely electric
or purely magnetic waves. As shown in the diagram, the waves
are structured so that the fields vibrate in directions that are
perpendicular both to each other and to the direction in which
the wave is moving. Unlike sound waves and water waves, electromagnetic waves are not vibrations of any physical medium.

A wave has a wavelength, which is the distance from one

crest to the next, and a frequency, which is the number of wave
crests that pass over a certain position in one unit of time. A
higher frequency means a shorter wavelength. Visible light is an
electromagnetic wave that has a frequency within a certain range,
with the blue and red ends of the rainbow differing by about a
factor of two. Visible light is only one small segment of the
vast electromagnetic spectrum, which includes such superficially
disparate phenomena as radio waves and x-rays.
Electromagnetic waves carry energy, as you can tell when you
use a microwave oven to heat your food.




Gravitational waves

The equivalence principle tells us that the gravitational field can,

at least locally, be considered to be an illusion: for a free-falling
observer, there is no field. This suggests the possibility that
waves in the gravitational field might also be illusions. In fact,
Einstein tried to publish a paper in 1936 making exactly this
claim. The paper was wrong, and was rejected based on the report of the Physical Reviews anonymous referee. (Einstein held
a grudge, and never submitted another paper to that journal.) In
nontechnical language, the important point is that although locally we can always consider a gravitational field to be fictitious,
that doesnt hold on larger scales. On larger scales, we can have
tidal forces (section 7.4, p. 100), which are objectively real; for
example, they can cause water to rise higher on a beach.

The figure shows what would happen if a strong enough gravitational wave passed through Albert Einsteins head. The effect
is to alternately squish and stretch him along two axes perpen-



dicular to the direction in which the wave is moving. Because

Einsteins head isnt perfectly flexible, it resists these distortions, and the result is frictional heating; just like electromagnetic
waves, gravitational waves transport energy. A more realistic way
of detecting a gravitational wave is to replace Einsteins head
with a kilometer-long vacuum tube, and monitor any stretching
or squishing of the tube using a laser; there is a ground-based
project called LIGO already in operation to try to do this, and a
European Space Agency project to launch a space-based system
called LISA.
Theres good news and bad news about detecting gravitational waves.
First the good news. In order to produce strong gravitational waves, you need large masses (such as stars) moving very
violently (preferably at speeds close to the speed of light). This
means that if astronomers can detect gravitational waves coming
from outer space, they can learn about rare awesomely apocalyptic processes, such as the collision of two black holes. Also,
gravitational waves do not interact very strongly with matter, so
the universe is essentially transparent to them. Our view of the
universe through the gravitational-wave spectrum would never be
blocked by clouds of gas or dust, as is often the case with electromagnetic waves. In particular, the very early universe (less
than about 100,000 years after the big bang) was full of hot, ionized gas, which was completely opaque to light, so gravitational
waves are probably the only way we could ever hope to see
the early universe directly. Direct detection of any gravitational
waves would also be a spectacular confirmation of one of general
relativitys most important predictions.
The bad news is the same as the good news. Processes violent
enough to produce strong gravitational waves are uncommon. If
such events happen, they happen rarely and typically at vast
distances from us. The weak interaction of gravitational waves
with matter also means that detectors made out of matter are
not very sensitive to them. Therefore its very difficult to get
such a detector to work well enough to give a definite blip.



Luckily direct detection is not the only possible way to verify

that gravitational waves exist. In 1974, Russell Hulse and Joseph
Taylor used the Arecibo radio telescope to discover a pair of neutron stars (see p. 139) orbiting around each other once every 7.75
hours. They found that this orbital period was gradually shortening, by about 68 nanoseconds with each revolution. Detailed
analysis of the radio signals allowed the stars masses and orbits
to be fully characterized. Because these are large masses moving
very violently, general relativity predicts that they will radiate
a huge amount of energy in the form of gravitational waves
7 1024 watts, or about a trillion times our worlds total production of electrical power. Because of this loss of energy, the stars
are predicted to be gradually spiraling inward toward one another, resulting in a quickening of their revolutions that matches
the observed speed-up to within a fraction of a percent. Because
of the excellent agreement with general relativity in this as well
as other observations of the system, physicists consider Hulse
and Taylors observations to be a very clearcut, albeit indirect,
confirmation of the existence of gravitational waves, and the two
physicists were awarded the Nobel Prize in 1993.

Part IV



Chapter 13

For dust thou art

In the sweat of thy face shalt thou eat bread, till thou
return unto the ground; for out of it wast thou taken:
for dust thou art, and unto dust shalt thou return.
Replace dust with atoms, and this verse from Genesis
tells a physics story in which matter is conserved, and atoms are
recycled as they pass through various forms. Dust also has
a technical meaning in relativity: a swarm of material objects,
all moving relative to each other at speeds that are small compared to c. A flock of galaxies, for example, would be dust
to a cosmologist, with the galaxies as the dust particles. In this
chapter we look at cosmological models in which the matter is
all in the form of dust, which until about 1998 was thought to
be a a realistic assumption.


The Einstein field equation

In John Wheelers formulation (ch. 10, p. 129), spacetime tells

matter how to move, and matter tells spacetime how to curve.
So far, weve applied these ideas to black holes, but to apply
them to cosmology we need to flesh out Wheelers description
a little more. Whereas a black hole is 100% empty space, the
cosmos is more of a uniform soup. Whether we look at a black



hole from a safe distance or are brave enough to dive in past

the event horizon, the effects we see are all due to the matter
contained in the singularity, which is at a distance from us. But
when we study the cosmos, we cant go outside and observe it
from far away as far as we know, there is no outside. Were
inside, like a fly in the soup.
This distinction between matter-here and matter-over-there is
directly coded into the Einstein field equation, which Ill describe
here in words and pictures rather than mathematical symbols.

In figure 1, someone releases a bag of marbles near the earth,

in a spherical cloud that is initially at rest and at some distance from our planet. As in the real-world example of Comet
Shoemaker-Levy (section 7.4, p. 100), the cloud stretches out due
to tidal stresses; the marbles that start out closer to the earth
experience stronger gravity and accelerate more quickly than the
ones farther away. But this is not the only effect on the shape of



the cloud. The whole cloud is angling in along a cone whose tip
is at the center of the earth, and therefore the cloud simultaneously gets narrower in the transverse direction. The result turns
out to be that although the cloud is distorted into an ellipse, its
volume stays exactly the same.
The cloud of marbles in figure 2 has been released so that
it encompasses the earth. This is gravity due to a mass that is
present, not absent. The cloud contracts in all three dimensions.
Its volume gets smaller, and its rate of contraction gets faster
as time goes on, because the marbles are accelerating, not just
moving at constant speed.
Although these pictures are presented in terms of a cloud of
physical particles, their only role was to free-fall along geodesic
paths and let us visualize that set of geodesics. These worldlines are initially parallel, because the marbles start at rest, but
as time goes on they start to converge. When lines start out
parallel but later start to converge or diverge, were dealing with
curvature (recall section 9.3, 122). What figures 1 and 2 really
show us, then, is a distinction between two types of curvature:
tidal curvature caused by matter-over-there as distinguished from
non-tidal curvature caused by matter-right-here.
The main idea of the Einstein field equation, translated into
words, is that it relates the non-tidal curvature to the matterright-here. Ordinary matter with positive mass-energy creates
non-tidal curvature that is positive, meaning that a cloud tends
to pull together. If its initially at rest, it contracts faster and
faster. If its initially expanding, its expansion slows down; it
may or may not reach a point where it stops and recontracts.
If there is no matter-right-here (other than that of the marbles themselves, which we assume to be negligible), we have only
tidal curvature. A common reason for tidal curvature is that
there is some matter-over-there, as in figure 1. This is not the
only possibility, however. For example, gravitational waves are
also a type of tidal curvature; perhaps you noticed the similarity
between figure 1 and the illustration on p. 160 of gravitational
waves stretching and squishing Einsteins head.



How do we measure matter for the purposes of the field

equation? Roughly speaking, we should use mass-energy, but in
fact the pressure of the matter also comes into play in a mathematical way that is too complex to describe here. It turns out,
however, that this pressure effect can be safely ignored if the
relative speeds of the matter particles are small compared to c,
which is part of what we mean by the definition of dust.



Many of the most fundamental laws of physics we have are conservation laws, which say that stuff cant just appear or disappear. For example, lets say in a cosmological context that stuff
means galaxies. Of course there is no specific law of conservation of galaxies, but counting galaxies is easy visually, and if an
entire galaxy did spontaneously wink out of existence (scary!), it
would certainly violate conservation laws such as conservation of

The figure shows a simple way of visualizing this on a space-



time diagram. Galaxies are entering and exiting the stage, but
that doesnt mean theyre being spontaneously created or destroyed. In fact we can verify the conservation of galaxies by
counting how many world-lines enter the rectangular figure and
how many exit. Seven come in (six through the bottom and
one through the left side) and seven go out (six out the top and
one out the right side). Because 7 = 7, we have conservation of
This counting technique is in fact a mathematically precise
and complete way of defining some of the fundamental conservation laws of physics, such as conservation of electrical charge.
When you use a laptop computer, you arent using up the charge
of the battery, although people may say it that way in popular speech. The same electrons stay in the laptop computer the
whole time; they only get shuffled around, not created or used
up. If we imagine hypothetically the physically impossible situation in which one of the electrons gets used up, it would be like a
situation in which one of the world-lines simply terminated in the
middle of the picture. That world-line would have entered the
picture but never exited, and the mismatch in the two counts
is how we would detect the violation (and presumably get the
Nobel prize for it).
I dont know of a nonmathematical argument to prove it,
but conservation of energy-momentum is directly implied by the
Einstein field equation; without it, the equation breaks down
and becomes mathematical nonsense. This is a firm prediction
of general relativity, so if any observation ever does disprove conservation of energy-momentum, even by some tiny amount, then
general relativity will have been disproved as well.
Conservation of energy-momentum is similar to the other examples above, but with the difference that energy-momentum
is a vector, not a scalar (section 4.3, p. 65, and section 5.2,
p. 72). This leads to an important loophole in general relativitys conservation of energy-momentum. If we draw our rectangular diagram too big, then we have a problem verifying that the
energy-momentum vectors coming into the box add up to the



same amount as what emerges. This is because if we want to

compare two vectors, we have to do it using parallel transport
(9.6, p. 126). But when spacetime is curved, parallel transport is
ambiguous. Therefore the conservation law only holds if we draw
a box small enough so that the curvature of spacetime doesnt
matter. In other words, conservation of energy-momentum in
general relativity is only a local conservation law.


Change or changelessness?

Now we have two tools: the Einstein field equation and conservation of energy-momentum. When Einstein invented these in
1915, scientists had a preexisting preference for uniformitarianism, which was the idea that the universe had always existed,
would always exist, and would, on the average, always look the
same. Uniformitarianism was a reaction against catastrophism,
a natural history dominated by dramatic events such as massive volcanic eruptions. The debate between the two schools of
thought had revolved around scientific questions such as whether
Californias spectacular Yosemite Valley had been formed by a
sudden subsidence or slow glacial erosion. The subtext was a suspicion that catastrophism was an attempt to shoehorn Biblical
events such as Noahs flood into the geological record.
The Einstein field equation doesnt allow for a static cosmos
with dust as its contents. Suppose that we had such a universe,
and consider a small region within it, containing a sampling of
galaxies (dust particles). The definition of dust tells us that
the galaxies are moving at nonrelativistic speeds relative to their
neighbors, so we can adopt a frame of reference tied to one of
the galaxies, and in this frame the speeds of the others will also
be approximately zero, as in spacetime diagram 1 in the figure.
Weve picked out a set of particles, initially at rest, very much
like one of the clouds of marbles in the figures on p. 166. But unlike the marbles, our galaxies have a significant amount of mass
of their own. Thats matter-right-here, so the cloud will start to



shrink faster and faster, as shown in figure 2. But this is a contradiction to our assumption that the universe was unchanging.
What general relativity does allow is a universe that is contracting, expanding (figure 3), or expanding and then recontracting
(figure 4).

Einstein realized this early on, but the idea of a contracting or

expanding universe had never been suggested on any theoretical
or empirical grounds, so he rejected it and focused his ingenuity
on fixing what he saw as a problem with his theory. It was not
until the 1930s that evidence began to appear that the universe
actually was expanding (section 3.10, p. 58). The theory was
worked out in full by the Russian physicist Alexander Friedmann,
and the Belgian priest Georges Lematre proposed that if one
extrapolated backward, there must have been a time when all of
the matter in the universe had emerged from a cosmic egg
what we now call the big bang.
Einsten hated it, and told Lematre, Your calculations are
correct, but your physics is abominable. In 1948, the British
astronomers Bondi, Gold, and Hoyle came up with an alternative, shown in figure 5, called the steady-state model. According
to this idea, the universe expands, and everything gets farther
apart, but at the same time atoms are spontaneously springing
into existence, so that on the average, the density of matter stays
the same. Thus although the universe isnt static (unmoving), it
looks the same at all eras of history. One problem with the theory
was that the proposed creation of matter is exactly the kind of
nonconservation of energy-momentum that is forbidden by general relativity. Although Hoyle went on trying to get around this
and other difficulties with the theory, it was eventually conclusively disproved (see section 13.6, p. 176).




Homogeneity and isotropy

The night sky looks like a bowl: the eye has no way of judging
depth. But once astronomers such as Henrietta Swan Leavitt
had developed the appropriate techniques, it became possible
to map out the structure of the universe on large scales the
scales on which we could consider galaxies as the dust grains
in our models. As surveys of the sky became more sensitive,
these maps were able to cover larger and larger distance scales,
like a cosmic zoom outward. Structure is evident on these maps,
including black empty spaces like the holes in Swiss cheese and
stringy structures like spiderwebs. These voids and filaments are
believed to have been formed by gravitational collapse that occurred during and despite cosmological expansion. But on the
very largest scales, beyond a few hundred million light years, we
no longer see any structure. The map looks like an undifferentiated, random scattering of galaxies.
It is therefore natural to simplify our cosmological models by
making the approximation that the universe is homogeneous on
the average, i.e., no region looks different from any other region.
Furthermore, it does not appear that the universe has any special axis or preferred direction, such as an axis of rotation or a
special direction with which spiral galaxies like to align themselves. This additional symmetry is called isotropy. The figures
on the facing page show visual examples that demonstrate these
two symmetries.






Hubbles law

The spacetime diagram in figure 1 on the facing page shows an

expanding universe, simplified by focusing on three galaxies that
happen to start out equally spaced along a line. This is the frame
of an observer in the middle galaxy, B. Galaxies A and C each
start out one unit of distance away from B, and over the course
of the time shown, they end up two units away.
Why do distances AB and BC both have to grow by the
same amount over the same time interval? They dont, and in
reality they probably wouldnt, if we were actually talking about
three nearby galaxies. But consider this instead as a cartoon
representing the structure of the universe on the largest scales.
Then by homogeneity, both of these distances have to expand at
the same rate.
Figure 2 shows the same situation in a frame where galaxy
A is at rest. Initially, observer Alice in galaxy A sees B as being
one unit away, and C as lying at a distance of two units. These
distances of 1 and 2 double to 2 and 4. Alice says that B is
moving away from her at some speed, and C is fleeing at double
that speed. (Of course Alice wont live long enough to watch
the process play out. She will actually measure these speeds
using Doppler shifts the technique pioneered by Slipher.) The
velocity of recession v is proportional to the distance d.
This proportionality is known as Hubbles law, after Edwin
Hubble, who worked at the Mount Wilson observatory near Los
Angeles. The constant of proportionality is called the Hubble
constant, H, so that we have v = Hd. Suppose for the moment
that, as implied by the diagrams, the motion of each galaxy is
inertial; gravity is negligible. Then by using the grade-school
equation (speed) = (distance)/(time), we find that if we extrapolate the expansion backward in time, there was a time equal to
1/H in the past when all three galaxies were in the same place.
That is, the Hubble constant can be interpreted as the inverse of
the age of the universe, provided that the expansion has been at
a constant rate.






The cosmic microwave background

Despite scientists initial reluctance to accept the big bang model,

many independent threads of evidence point to it. For example,
our universe has quite a bit of hydrogen-2, also known as deuterium. This is an isotope of hydrogen in which the nucleus has
one proton and one neutron. The present-day universe has physical processes in the cores of stars that steadily destroy deuterium,
but no processes that produce it. If the universe had been infinitely old and always in its present state, as in the steady-state
model, all of the deuterium should have long since been eliminated.

The steady-state model finally went to its grave in 1964 when

Arno Penzias and Robert Wilson at Bell Labs in New Jersey
found an annoying source of background radiation in the signals
picked up by the microwave antenna, shown in the photo, which
they had hoped to use for satellite communications and radio
astronomy. They tried to fix their apparently broken instrument
by rousting out the pigeons that had been nesting in it, and
cleaning out the droppings. The signal was still there. Was it
human-generated, like radio waves from power lines or cars spark
plugs? But that type of noise would typically change in intensity
between day and night. Not only did the signal fail to vary with a



24-hour periodicity, but it also didnt show the telltale variation

over 23.93 hours (a sidereal day, the time it takes the earth to
spin on its axis) that is characteristic of emissions from a source
at one location in the sky.
Eventually they became convinced that these radio waves
were coming from outside our galaxy, and uniformly from all directions in the sky. Communication with the physics community
at nearby Princeton led them to realize that they had detected
the afterglow of the big bang, Doppler shifted from the visible
spectrum down into radio. The big bang model accounts not only
for the existence of the radiation but also for its detailed characteristics, such as the variation of the intensity with wavelength.
The steady-state model is unable to account for these facts. The
radiation is referred to as the cosmic microwave background, or


A hot big bang

These days the study of the CMB is an entire scientific field of its
own, as technical and sophisticated as the study of wine, ballet,
plumbing, or rock climbing. Without delving into the details too
deeply, we can pick out some straightforward ideas. Its a glow,
and hot things glow, so we suspect that the universe used to be
hot. The temperature scale we use for this is the Kelvin scale,
whose zero is absolute zero, the temperature at which all random
motion ceases completely. A Kelvin degree (K) is the same size
as a Celsius degree, so the Kelvin and Celsius scales are the same
except for an additive constant. Everyday temperatures that we
experience on our planet are in the ballpark of 300 K.
The mixture of wavelengths emitted by a glowing object depends on its temperature. Right now, your body is emitting
infrared radiation, which means light with a wavelength a little
too long to be visible without infrared goggles. A hot poker glows
red. If you heat it more and more, the mixture of wavelengths
it emits shifts toward lower and lower wavelengths. Presently,



the CMB has a mixture of wavelengths that is typical of a temperature of only 2.7 K, but these wavelengths are not the ones
originally emitted. General relativity predicts that as the universe expands, and galaxies get farther apart, the space between
them also expands. Because of this expansion, when a light wave
travels across the universe for billions of years, its wavelength
Since most of the normal matter in the universe is hydrogen,
we expect that this light was originally emitted by hot hydrogen
gas at a temperature of about 3000 K, which is the highest at
which hydrogen is transparent; at higher temperatures, the electrons are stripped off of the atoms, and the gas becomes opaque.
So the CMB is a relic of an era in which the universe had a temperature of about 3000 K. When originally emitted, the CMB
has a yellowish-orange color. The drop of a factor of about 1000
in temperature between then and now indicates that the wavelengths have expanded by 1000 times, and therefore that cosmological expansion since that era amounts to a factor of 1000.
We are led to cosmological models in which the big bang was
hot. Presumably it was even hotter at times before the CMB was
emitted. Support for this extrapolation comes from the mixture
of chemical elements and isotopes of those elements that we observe in the present universe. We noted on p. 176 that deuterium
(hydrogen-2) is currently being depleted by nuclear reactions in
stars. That means that the universes present supply of deuterium must have originated from nuclear fusion reactions in the
first few minutes after the big bang, when the temperature was
millions of degrees kelvin.


A singularity at the big bang

It seems as though the universe keeps getting hotter and denser

as we extrapolate back in time closer and closer to the big bang.
When Friedman first constructed isotropic and homogeneous cosmological models using the Einstein field equations, he always got



results in which the temperature and density blew up to infinity

at the big bang. This would indicate a singularity, similar to,
but different from, a black hole singularity. Cosmologists were
at first skeptical as to whether these singularities were real. The
assumption of homogeneity was after all only an approximation.
They expected something more like the situation shown in the
figure, where instead of a big bang we have a big near miss. This
expectation was natural based on their knowledge of the universe, because in general astronomical objects are tiny compared
to the distances between them, so they make small targets, and
collisions are extremely rare.

But in the 1970s, relativist Stephen Hawking (b. 1942) proved

the Hawking singularity theorem, which tells us that there was
a big bang singularity provided that three assumptions are satisfied: (1) general relativity and the Einstein field equation are
valid, (2) matter behaves in a reasonable way by obeying certain energy conditions (p. 134), and (3) present-day conditions



in the universe fall a within certain range of parameters. (A similar theorem was proved by Penrose for black hole singularities.)
Condition 3 has been verified definitively through astronomical
observations. Condition 2 is believed to have been valid in the
early universe. Condition 1 is expected to hold up to certain
known, very high densities and temperatures, at which quantum
mechanics is expected to invalidate general relativity.

Like a black hole singularity, the big bang singularity is spacelike. A black hole singularity becomes an inevitable part of the
future light cone of an observer who passes inside the event horizon, but other observers never come in contact with it. The big
bang singularity, on the other hand, lies in the past light cone of
every observer.
The figures show two possible scenarios according to Friedmann. In figure 1, we see the world-lines of several objects in
a universe that has enough mass in it so that gravitational attraction gradually slows down the expansion of the universe and



brings it to a halt. After that, the universe recontracts and

ends up going out in style, in a fiery implosion called, naturally
enough, a big crunch. Figure 2 shows the same cosmology as a
Penrose diagram. Because Penrose diagrams dont try to show
times and distances to scale, the expansion and recontraction are
not shown. What the diagram does show accurately is that in
this type of universe, space is finite, and it wraps around on itself.
In figures 3 and 4 we see a possibility in which the universe
does not contain enough mass to cause it to recontract. Its expansion is analogous to the motion of a projectile that has been
shot up from the earths surface at greater than escape velocity,
so it will never fall back down. This universe is infinite both in
space and in time.
The Penrose diagram in figure 4 looks exactly like the one
in figure 1 on 147, for the future half of the flat spacetime of
special relativity. The past half is missing, because there was no
time before the big bang. These two spacetimes are not actually
the same one is curved and the other is flat. But Penrose
diagrams arent designed to show every aspect of curvature. The
two spacetimes do have the same structure in terms of light cones,
and cause and effect.
Figures 2 and 4 accurately show that the big bang was not
an explosion that happened at one point in space. The big bang
happened everywhere at once. In the case where the universe is
infinite, figure 4, it has always been infinite and always will be.
The triangular shape of figure 4 makes it look as though everything in the universe is getting closer together over time, which
would be a contraction rather than an expansion. But remember,
Penrose diagrams arent supposed to show distances on a consistent scale. Similarly, the parallel painted lines on the highway in
the photo on p. 146 look like theyre getting closer together, but
actually theyre not; even if the people who painted the lines had
decided to make them get gradually farther and farther apart,
they would still appear to the eye to converge at a vanishing
It may seem strange to imagine an infinite universe that is



expanding. If infinity is already bigger than anything, how can

it get bigger than that? This is similar to a famous paradox
proposed by the mathematician Hilbert. Suppose we have a hotel
with infinitely many rooms, and each of them is occupied by one
of the infinitely many dentists attending a convention. While
ToothCon is still going on, a soybean farmers convention begins,
and there are also infinitely many farmers attending. If the hotel
is the only one in town, and its already full of dentists, where
can the farmers stay? We can accomodate them in the same
hotel. All we have to do is move the dentist from room 1 into
room 2, and similarly from room 2 to 4, 3 to 6, and so on. Now
only the even-numbered rooms are occupied, and the farmers can
check in to the odd ones. Similarly, we can have a universe with
an infinite supply of cubic centimeters of space, but still have
it make sense to talk about doubling its volume over a certain
period of time.

Regardless of whether the universe is finite or infinite, the part

of it that we can presently observe is finite, because its confined
to our past light cone. This is called the observable universe.
The edge of the observable universe is at a distance from us such
that information traveling at the speed of light would have just
arrived here today if it had been emitted immediately after the
big bang.

A common solecism is to fail to distinguish between the universe and the observable universe, and to use the same word
universe for both. For example, there are about 100 billion
galaxies in the observable universe, but people will sometimes
say that there are this many galaxies in the universe. This
is incorrect. The universe may be infinite, and in that case the
number of galaxies in it is presumably infinite as well.




A cosmic calendar

Ive been blithely referring to ideas like the age of the universe
and the time at which the CMB was emitted. When we use words
like age and time, we need to be careful, because relativity
tells us that time is not absolute. For example, the age of the
universe is currently estimated at 14 billion years, but whose
clock would we use to measure that time? It makes a difference
what state of motion the clock is in.
But because the universe has been evolving over time, we have
a natural way of getting around this ambiguity. For example,
the average temperature of the universe (meaning the average
temperature of all the matter and light in it) has gone down
over time. In the figure on p. 184, we have a spacetime diagram
that shows this cooling process. All three thermometers are at
rest relative to the average motion of the nearby galaxies. By
homogeneity, all three thermometers measure the same process
of cooling, which happens at the same rate according to a clock
attached to the thermometer. The result of all this is that we
get a preferred notion of simultaneity, indicated by the horizontal
dashed line in the figure. We are free to pick some other notion
of simultaneity, as suggested by the slanted line, but this is less
natural and convenient.
This whole process is similar to how we can estimate the time
of year by looking out the window. If theres snow on the ground
in my front yard, its probably December, not July.



Chapter 14

Dark matter and dark

Some say the world will end in fire,
Some say in ice.
Robert Frost
How will the world end? Nuclear war? Environmental ruin?
Zombie apocalypse? Surprisingly, science has an easier time answering this question for the universe as a whole than for our own
planet, which we can so easily touch, see, and feel. How, then,
will the universe end its days?


A difficult census

If gravity didnt exist, then the answer would clearly be neither

fire nor ice but dilution. Everything is moving farther apart, and
as time goes on it gets more spread out. The lumps of matter
are separated by more and more empty space. If the expansion
continues indefinitely, and at a constant rate, then the universe
will eventually get to be like the thinnest soup imaginable.
Gravity does exist. Suppose, as was believed for most of the
20th century, that the universes matter consists mostly of ordinary atoms (not, say, exotic subatomic particles), which move


nonrelativistic speeds relative to other nearby matter. It then
qualifies as the dust of ch. 13. The Einstein field equation tells
us that the expansion is not at a constant rate. Since dust satisfies certain energy conditions (p. 134), the result is as we would
expect according to Newtons description of gravity as an attractive force. The expansion decelerates. We then have the two
possible scenarios described in section 13.8, p. 178. Either the
universe recollapses into a big crunch, or it goes on expanding
forever. To figure out which of these would happen, astronomers
devoted a great deal of effort in the 20th century to trying to measure the average density of the matter in the cosmos. Essentially
they were trying to take a census of the universes contents.


Dark matter

In our own solar system, the sun accounts for 99.8% of the mass.
(The earths contribution is only a few ten thousandths of a percent.) Stars are bright objects, so wed imagine it would be easy
to peer through a telescope and map out almost all of the universes mass.
If only it were so simple. As discovered by Vera Rubin in the
1970s, stars orbit within their galaxies at high speeds, and if the
only mass present was that of the other stars, there would not
be enough gravitational attraction to keep them in orbit. They
would be ejected, like the sparks flying off of the grinding wheel
in the photo. Similar observations apply at the scale of clusters of
galaxies. There must be a large amount of nonluminous matter
to hold these structures together.
Now the weirdness starts. We have multiple independent lines
of evidence telling us that this mysterious matter cant be made
out of atoms:
1. If the universe contained this much matter in the form of
atoms, then nuclear fusion reactions in the first few minutes
after the big bang would have proceeded more efficiently,



and the universe would contain a higher proportion than is

actually observed of elements such as helium.
2. The CMB is very nearly uniform, but does contain tiny
irregularities, which are clues as to what happened in the
early universe. The existence of non-atomic matter is required in order to make models that reproduce these fluctuations.
3. In merging galaxy clusters, it has been observed that the
gravitational forces dont point in the direction they should
if they were being produced by the clouds of gas made of
normal, atomic matter.
Because we dont know what this exotic matter is, we just
call it dark matter. Our best guess is that it consists of some
kind of subatomic particle, similar to the neutrino (which we met
on p. 89), that doesnt interact strongly with ordinary matter.
(We know that it cant actually be neutrinos, though, because
we can detect neutrinos, and there arent enough of them flying
When the Large Hadron Collider came online in 2010, it was
hoped that it would create a slew of new particles, some of which
might be the same as the ones in dark matter. That didnt


happen, so we remain ignorant. A number of experiments, buried
in deep mineshafts to shield against background noise, are under
way to detect dark matter directly. As of 2013, the most sensitive
experiment has given null results.1


Dark energy

As if this wasnt weird enough, astronomers determined in 1998

that the acceleration of the universe was accelerating rather than
decelerating. The original evidence came from the Doppler shifts
of supernovae (exploding stars) in distant galaxies. Because the
light from such galaxies has taken billions of years to reach us, it
gives us a window into the past, and allows us to determine how
fast the universe used to be expanding. Unexpectedly, the data
showed that the universe used to be expanding more slowly, and
has now sped up.
The statistical quality of the supernova data was not very
solid, and many people, including me, were skeptical. Extraordinary claims require extraordinary evidence. But since then, the
acceleration has been confirmed by two other independent methods (analysis of CMB fluctuations, and a method called baryon
acoustic oscillations).
The Einstein field equation tells us that this cannot happen
as a result of the gravitational fields of ordinary matter or dark
matter. These forms of matter obey the energy conditions
roughly speaking, their mass is positive, rather than negative
and therefore they create a gravitational attraction that would
decelerate the expansion of the universe. The acceleration must
be caused by something else, which we refer to as dark energy.
According to current estimates, the universe is about 68% dark
energy, 27% dark matter, and 5% matter made of atoms.
Unlike dark matter and ordinary matter, dark energy is believed to be nearly uniformly distributed across the universe; if it
were concentrated in galaxies, then it would cause those galaxies




to fly apart. In our current models, dark energy is described as

a kind of energy that is automatically built in to the structure of
empty space. As the universe expands, more space is produced,
and this leads to the production of more dark energy, which fuels further expansion. (Surprisingly, the careful mathematical
analysis of this process shows that it does not violate the local
conservation of energy-momentum described in section 13.2.) In
the early universe, the gravitational fields were dominated first by
light and later by matter, but we have now entered the universes
final era, in which the dominant effect is that of dark energy.
The figure below shows the Penrose diagram for a universe
that, like ours, ends up dominated by dark energy.

In closing, its worth remarking on the fact that in relativity and cosmology, as is often the case in the sciences, we learn
definite answers to questions that most people would imagine
science could not answer, while conversely we know very little
about things that laypeople would think were elementary and
well established. We know all about the detailed history of the
universe and its ultimate fate, questions that would traditionally
have been relegated to myths or religion. At the same time, we
turn out to know nothing at all about what 95% of the universe
is made of.


Photo credits
Cover Clock face: Wikimedia commons user Lexlexlex, retouched by user
Cover Ladybug: Redrawn from a photo by Wikimedia Commons user Gilles San Martin, CC-BY-SA.
14 Runner: Redrawn from
photos by Muybridge, 1887.
17 Car: Redrawn from a photo by Wikimedia Commons user Auregann, CC-BY-SA. 17 Cow: Redrawn from a
photo by Wikimedia Commons user Cgoodwin, CC-BY-SA.
18 Plane:
Redrawn from a photo by Wikimedia Commons user Andy Mitchell, CCBY-SA. 22 Cow and car: As for the figure on p. 17.
35 Muon storage ring at CERN: (c) 1974 by CERN; used here under the U.S. fair use
36 Plane: Redrawn from a photo by Wikimedia Commons
user Andy Mitchell, CC-BY-SA. 36 Rotating earth: NASA image, public domain; animated gif by Wikimedia Commons user Marvel, CC-BY-SA.
45 Duck-rabbit illusion: J. Jastrow, Popular Science Monthly, 1899.
Galaxy: NASA/STScI/ESA, public domain.
56 Cadillac: Redrawn from
a photo by That Hartford Guy, CC-BY-SA. 72 Astronaut: NASA. 73
Fireworks: Kristina Servant, CC-BY.
73 Sprinters: Darren Wilkinson,
CC-BY-SA. 83 PET scan: NIH, public domain. 99 Football player and
old lady: Hazel Abaya.
101 Impact sites on Jupiter: NASA, public domain. 101 Comet breaking up: NASA/ESA, public domain. 104 M100:
European Southern Observatory, CC-BY-SA. 104 Saturn: NASA, public domain.
104 Girl: Georges Hebert, Leducation Physique feminine,
1921, public domain.
105 Airplane: From a photo by Major James Skitt
Matthews (Canada), 1930, public domain. 105 Artificial horizon: NASA,
public domain.
106 M100: European Southern Observatory, CC-BY-SA.
106 Ladybug: Redrawn from a photo by Wikimedia Commons user Gilles
San Martin, CC-BY-SA. 108 Hawking aboard Vomit Comet: Public domain product of NASA. 111 Standing woman: Redrawn from work by
William Crochot, CC-BY-SA. 111 Jumping man: Redrawn by the author
from photos by Eadweard Muybridge,ca. 1880.
113 Pound and Rebka
photo: Harvard University. I presume this photo to be in the public domain,
since it is unlikely to have had its copyright renewed. 119 Standing woman:
Redrawn from work by William Crochot, CC-BY-SA. 120 Ladybugs: Redrawn from a photo by Wikimedia Commons user Gilles San Martin, CCBY-SA. 120 M100: European Southern Observatory, CC-BY-SA. 122
Galaxy and star: Hubble Space Telescope. Hubble material is copyrightfree and may be freely used as in the public domain without fee, on the
condition that NASA and ESA is credited as the source of the material.
The material was created for NASA by STScI under Contract NAS5-26555
and for ESA by the Hubble European Space Agency Information Centre.
122 Saturn: Public domain, NASA. 122 Vesta (made spherical): NASA,
public domain.
122 Human figures: Line art by B. Crowell, CC-BY-SA
licensed. Based on a photo by Richard Peter/Deutsche Fotothek, CC-BY-SA


licensed.. 124 Eclipse: 1919, public domain. 132 Hovercraft: Copyright

status unknown; used here under the U.S. fair use doctrine. 133 Upsidaisium: Copyright Jay Ward Productions; used here under the U.S. fair use
136 Crushed cars: Wikimedia Commons user IFCAR, public domain.
138 Sun: NASA/SDO (AIA), public domain.
138 Star
field: G. Bacon (NASA/ESA/STScI), public domain. 141 Satellite: From
a public-domain artists conception of a GPS satellite, product of NASA.
141 Woman on earth: Redrawn from work by William Crochot, CC-BY-SA.
141 Astronauts: Line art by B. Crowell, CC-BY-SA licensed. Based on
a photo by Richard Peter/Deutsche Fotothek, CC-BY-SA licensed.
Accretion: S. Gezari, J. Guillochon, STScI/NASA, public domain.
Lensing: Wikimedia Commons user Alain r, CC-BY-SA. 146 Rear-view
mirror, main landscape: Bradley Gordon, CC-BY.
146 Rear-view mirror, landscape in mirror: Benjamin D. Esham, CC-BY-SA.
148 Landforms for Goode homolosine: Wikimedia Commons user Strebe, CC-BY-SA.
148 Mercator projection: Wikimedia Commons user Eric Gaba, CC-BYSA. 148 Grid for Goode homolosine: Wikimedia Commons user Strebe,
CC-BY-SA. 160 Einstein: Oren Jack Turner, 1947, public domain in the
U.S. 168 Galaxies: Ville Koistinen, CC-BY-SA. 168 Galaxies: Ville
Koistinen, CC-BY-SA. 172 Hubble: 1920, public domain.
172 Leavitt:
public domain.
172 Slipher: 1905, public domain.
173 Galaxy: Ville
Koistinen, CC-BY-SA. 175 Galaxies: Ville Koistinen, CC-BY-SA. 176
Penzias-Wilson antenna: NASA, public domain.


E = mc2 , see mass-energy equiv- boost, see Lorentz transformaalence
Bridgman, Percy Williams, 63
maximum speed of cause and
causality, 57
effect, 38
Chandrasekhar, Subrahmanyan,
near a black hole, 151
no acceleration past, 51
circular motion, 18
experiments, 52
CMB, see cosmic microwave backspeed of light, 39
tachyons, 88
conservation laws
acceleration, 95
energy, 74
action at a distance, 155
momentum, 73
antimatter, 82
general relativity, 168
energy-momentum, 169
in spacetime diagrams, 61
only local, 170
Aristotle, 11
special relativity
Bell, Jocelyn, 139
mass-energy, 82
big bang
coordinates, 16
singularity, 179
Copernicus, 98
black hole
correspondence principle, 31
Cygnus X-1, 143
combination of velocities, 41
detection, 142
Lorentz transformation, 50
event horizon, 140
cosmic microwave background,
Penrose diagram, 146
singularity, 139
supermassive, 143
big bang, 171

static, 170
steady state, 171
disproved by CMB, 176
Cygnus X-1, 142
dark energy, 188
dark matter, 186
deuterium, 176
Doppler shift
cosmological, 174
energy, 77
frequency, 58
dust (relativistic usage), 165

force, 99
gravity is not, 104
frame of reference, 16
Galilean relativity, 16
general relativity, 104
no preferred frame, 16
only local in general relativity, 109
free fall
acceleration, 95
universality, 93
equivalence principle, 103
experimental tests, 96
explained by Newtons laws,
friction, 12
Friedmann, Alexander, 171, 178

Eotvos, 96
Einstein field equation, 165
electric field, 158
electromagnetic wave, 158
spectrum, 159
ellipse, 98
Galilean relativity, 11
energy, 74
Galilean transformation, 22
kinetic, 74
Galileo, 11
relativistic, see energy-momentum
conflict with the Church, 23
free fall, see free fall
energy conditions, 134
violated by dark energy, 134 gamma ray, 82
equals 1 in relativistic units, 42 general relativity
Einstein field equation, 165
equivalence principle, 103
tests of
defined, 104
deflection of starlight, 123
event, 28
Gravity Probe B, 128
event horizon
Hulse-Taylor pulsar, 162
black hole, 140
orbit of Mercury, 150
verbal summary, 129
falsifiability, 44
generator, 158
field, 155
GPS, 33
electric, 158
magnetic, 158

apparent weightlessness, 107
as curvature of spacetime,
no shielding, 131
not a force, 104
not repulsive for ordinary
matter, 132
wave, see gravitational wave
greatest time, see time, principle of greatest
Hafele-Keating experiment, 30
homogeneity, 172
Hoyle, Fred, 171
Hubble constant, 174
Hubbles law, 174
Hubble, Edwin, 174
Hulse-Taylor pulsar, 162
induction, 158
inertia, 18
Newtons first law, 100
isotropy, 172

length contraction, 55
not what you see, 55
light, 75
electromagnetic wave model,
see electromagnetic wave
gamma ray, 82
models, 75
light cone, 42
tipping, 143
lightlike, see vector
Lorentz contraction, see length
Lorentz transformation, 48
correspondence principle, 50
doesnt preserve simultaneity, 49
preserves area, 61
Lorentz, Hendrik, 48
lune, 121

magnetic field, 158

mass, 71
kilogram unit, 72
operational definition, 71
mass-energy equivalence, 80
satellites, 14
correspondence principle, 81
Shoemaker-Levy 9 impact,
finite strength, 134
no nongravitating, 130
Kelvin scale, 177
maximal time, principle of, see
Kepler, 98
time, principle of greatKuhn, Thomas, 45
Mercury, 150
Large Hadron Collider, 187
momentum, 72
Lavoisier, 82
Leavitt, Henrietta Swan, 172
relativistic, see energy-momentum
Lematre, Georges, 171

muon, 35


Sagittarius A*, 143

neutrino, 89, 187
Galilean, 64
neutron star, 139
special relativity, 66
Newton, Isaac, 30, 98
laws of motion, 100
absolute in Galilean relativity, 28
observable universe, 182
relative in special relativoperationalism, 63
ity, 40
paradigm shift, 45
big bang, 179
parallel transport, 126
black hole, 139
Pauli, Wolfgang, 71
Sokal, Alan, 44
Penrose diagram, 146
solar system
black hole, 146
orbits, 96
cosmology, 181
spacelike, see vector
universe with dark energy, spacetime, 28
spacetime diagram, 28
Penzias, Arno, 176
special relativity, 31
PET scan, 83
speed of light, see c
planets, see solar system
Popper, Karl, 44
evolution and death, 137
positron, 82
neutron star, 139
Pound-Rebka experiment, 113
pulsar, 139
principle of greatesr time, see
white dwarf, 138
time, principle of great- static cosmology, 170
steady state, see cosmology
proper time, 67
tachyons, 88
pulsar, 139
temperature, 177
relativity, see Galilean relativ- tidal forces, 100, 167
ity, special relativity, gen-time
eral relativity
absolute in Galilean relativrevolutions, scientific, 44
ity, 30
Rossi-Hall experiment, 35
Rubin, Vera, 186
due to gravity, 110

due to motion, 55
wave, 155
gravitational, 160
experiments showing relaweightlessness, see gravity
tivity of
Wheeler, John, 129
airplanes, 31
white dwarf, 138
mountain-valley experiments,
Wilson, Robert, 176
subatomic particles, 35
principle of greatest
general relativity, 118
special relativity, 36
simultaneity, see simultaneity
timelike, see vector
transformer, 158
twin paradox
Galilean, 21
special relativity, 36
observable, 182
Galilean, 64
lightlike, 67
magnitude, 65
parallel transport, 126
spacelike, 67
timelike, 67
combination of
correspondence principle,
Galilean relativity, 19
special relativity, 40
relativistic, 39
ultrarelativistic, 53